<%BANNER%>

Understanding Variability in Reaching Movements Post Stroke

Permanent Link: http://ufdc.ufl.edu/UFE0042466/00001

Material Information

Title: Understanding Variability in Reaching Movements Post Stroke Non-Linear Dynamical Systems Perspective
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Sethi, Amit
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: kinematics, motor, stroke, upper, variability
Rehabilitation Science -- Dissertations, Academic -- UF
Genre: Rehabilitation Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Stroke is the leading cause of long-term disability in the United States and around the world (Duncan, 1995). Upper extremity impairment is one of the most frequent impairments after stroke (Gresham, et al., 1975). The damage to the motor system caused by the stroke results in imperfect motor control, often exhibited as atypical or stereotypical movement patterns. One hallmark of this dysfunctional motor system is the high variability present in several movement parameters, such as upper extremity joint range of motion, movement time, and peak velocity, when variability is conceptualized from the traditional motor control perspective (Cirstea & Levin, 2000). Movement variability under these traditional motor control theories is considered as undesirable noise in the motor output (Stergiou, Buzzi, Kurz & Heidel, 2004), and therefore error. However, contemporary motor control theories, such as dynamic systems theory, consider variability as an intrinsic characteristic of movement and plays an integral role in motor learning (Bernstein, 1967; (Kamm, Thelen, & Jensen, 1990). Developing variability is indicative of the development of greater functionality in the motor system. Variability in healthy biological systems reveals the inherent complexity of the system components and their functional interactions (Vaillancourt & Newell, 2002). Further, variability in various physiological systems not only explains the complexity of a healthy system but also reflects adaptability and flexibility to the system. It might then seem intuitive to refer the complexity of the system as, ?adaptive variability?. Lipsitz and Goldberger (1992) proposed a ?loss of complexity hypothesis? suggesting a decline in optimal complexity as a function of aging and disease. Because of the damage to motor neural networks from stroke, it would then seem intuitive that individuals with stroke might also exhibit reduced complexity or adaptive variability in upper extremity movement. If this assumption holds true, enhancing upper extremity adaptive variability might seem to be an important goal of upper extremity stroke rehabilitation. Therefore, the overall purpose of this dissertation is to examine the adaptive variability of movement in the upper extremity and understand its relationship to upper extremity motor performance post stroke. This dissertation comprises three studies. The first study investigates whether the adaptive variability of upper extremity joint kinematics is reduced post stroke. The second study examines some of the task variables that might augment upper extremity adaptive variability post stroke. The final study investigates the change of upper extremity adaptive variability in individuals with stroke following an intervention based upon Constraint Induced Movement Therapy.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amit Sethi.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Richards, Lorie G.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042466:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042466/00001

Material Information

Title: Understanding Variability in Reaching Movements Post Stroke Non-Linear Dynamical Systems Perspective
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Sethi, Amit
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: kinematics, motor, stroke, upper, variability
Rehabilitation Science -- Dissertations, Academic -- UF
Genre: Rehabilitation Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Stroke is the leading cause of long-term disability in the United States and around the world (Duncan, 1995). Upper extremity impairment is one of the most frequent impairments after stroke (Gresham, et al., 1975). The damage to the motor system caused by the stroke results in imperfect motor control, often exhibited as atypical or stereotypical movement patterns. One hallmark of this dysfunctional motor system is the high variability present in several movement parameters, such as upper extremity joint range of motion, movement time, and peak velocity, when variability is conceptualized from the traditional motor control perspective (Cirstea & Levin, 2000). Movement variability under these traditional motor control theories is considered as undesirable noise in the motor output (Stergiou, Buzzi, Kurz & Heidel, 2004), and therefore error. However, contemporary motor control theories, such as dynamic systems theory, consider variability as an intrinsic characteristic of movement and plays an integral role in motor learning (Bernstein, 1967; (Kamm, Thelen, & Jensen, 1990). Developing variability is indicative of the development of greater functionality in the motor system. Variability in healthy biological systems reveals the inherent complexity of the system components and their functional interactions (Vaillancourt & Newell, 2002). Further, variability in various physiological systems not only explains the complexity of a healthy system but also reflects adaptability and flexibility to the system. It might then seem intuitive to refer the complexity of the system as, ?adaptive variability?. Lipsitz and Goldberger (1992) proposed a ?loss of complexity hypothesis? suggesting a decline in optimal complexity as a function of aging and disease. Because of the damage to motor neural networks from stroke, it would then seem intuitive that individuals with stroke might also exhibit reduced complexity or adaptive variability in upper extremity movement. If this assumption holds true, enhancing upper extremity adaptive variability might seem to be an important goal of upper extremity stroke rehabilitation. Therefore, the overall purpose of this dissertation is to examine the adaptive variability of movement in the upper extremity and understand its relationship to upper extremity motor performance post stroke. This dissertation comprises three studies. The first study investigates whether the adaptive variability of upper extremity joint kinematics is reduced post stroke. The second study examines some of the task variables that might augment upper extremity adaptive variability post stroke. The final study investigates the change of upper extremity adaptive variability in individuals with stroke following an intervention based upon Constraint Induced Movement Therapy.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amit Sethi.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Richards, Lorie G.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042466:00001


This item has the following downloads:


Full Text

PAGE 1

1 UNDERSTANDING VARI A BILITY IN REACHING MOVEMENTS POST STROKE: NON LINEAR DYNAMICAL SYSTEMS PERSPECTIVE By AMIT SETHI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF TH E REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Amit Sethi

PAGE 3

3 To Mummy and Papa

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my heartfelt gratitude to my advisor Dr. Lorie Richards whose encouragement, guidance and support from the initial to the final level enabled me to develop an understanding of rehabilitation research. I am immensely thankful to Dr. Richards for providing the constant encouragement and unrelenting support to pursue a research endeavor of my interest. She always motivated me and helped to seek resources so that I could acquire the best possible knowledge. I cannot thank her enough for teaching me the critical skill of establishing a link between theory research and its application to our clien ts in health care. I am also indebted to the exceptional members on my committee Dr. Carolynn Patten, Dr. James Cauraugh and Dr. Nicholas Stergiou for their invaluable support, belief in my work and for inspiring the art of being a scientist. Dr. Patten provided the scientific rationale for utilizing the complex terminology of dynamical systems theory and inspired me to pursue research in this complex area of motor control. Dr. Cauraugh provided a good and strong foundation for conducting scientific rese arch. Dr. Stergiou deserves special thanks for being a special member of my committee from University of Nebraska, Omaha. Dr. Stergiou taught me to apply complex engineering techniques, which formed the backbone of my research and also guided me to interpr et the data in relation to population with impairments. I would also like to thank an earlier member of my committee, Dr. Steve Kautz who guided me to think beyond and examine the minute details of the data. I would also like to thank Dr. Samuel Wu, for as sistance with statistical analysis of the data. Many t hanks to Theresa McGuirk, biomedical engineer for all her tireless work and assistance with data collection and analyses and Tara Patterson a former student

PAGE 5

5 in the lab for her help in data collection. I would also like to acknowledge the members of the upper extremity initiative of Brain Rehabilitation Research Center who have taught me to be a team player. Sandy Davis deserves special thanks for supporting me in all the ups and downs experiences durin g my graduate education at U niversity of F lorida ( UF ) I want to thank all participants who volunteered to participate in my research studies and provided me an opportunity to advance the field of rehabilitation science with my research endeavors. None of my research pursuits would be possible without them. My doctoral education would not be a possibility without the strong infrastructure and resources provided by UF for graduate education. I am grateful to the UF Alumni Association for granting me the Al umni fellowship that funded my education for four years. I am also thankful to N ational I nstitute s H ealth and V eteran A ffairs Medical Center for their grant funding to Dr. Lorie Richards that provided me access to the data of the research participants. I a m also grateful to the research grant provided by College of Public Health and Health Professions, UF for providing me the support to conduct my last study. I am also grateful to the faculty and staff in the Rehabilitation Science Doctoral (RSD) Program an d Occupational Therapy Department who have been an essential support system throughout my doctoral studies. Special thanks go to fellow RSD graduate students and all my friends here in the US for their great friendships and making Gainesville my home awa y from home. I cannot thank enough to my engineering friends, Karthik, Vijay, Ankit, Priyank, and Mohsen who have always helped to crack any Matlab problem and being there for me at times when challenged by my c omputer programming skills

PAGE 6

6 None of this wo sister in law. I am immensely grateful to them for selflessly loving me and supporting my desire to pursue doctoral education. My parents were with me for a majority of time of the last two ye ars of my doctoral education and provided me the strength and determination to achieve my goals. I cannot forget my adorable nieces Aashka and Anahita for always bringing a smile to me. Finally, thank you God for giving me the courage and will to pursue m y dreams and for building everlasting friendships and relationships during my doctoral education.

PAGE 7

7 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 2 THEORETICAL B ACKGROUND AND SIGNIFICANCE ................................ ......... 19 Section 1: Upper Extremity (UE) Impairment and Movement Variability in Stroke .. 19 Stroke: Incidence and Consequences ................................ .............................. 19 Pathophysiological Basis of UE Impairment and Performance Post Stroke ..... 20 Motor Recovery Post Stroke and Movement Variability ................................ ... 21 Section 2: Theoretical Basis of Movement Variability ................................ ............. 21 Variability in Movement Past and Current Perspective ................................ .. 21 Traditional view of motor control and learning ................................ ............ 21 Linear measures of variability ................................ ................................ .... 25 Variance/standard deviation (SD) ................................ .............................. 25 Coefficient of variation (CV) ................................ ................................ ....... 26 Contemporary view of motor control and learning ................................ ...... 27 Understanding complexity of movement ................................ .................... 30 Adaptive variability and stroke ................................ ................................ ... 31 Underst anding complexity in UE movements using non linear tools .......... 32 Approximate entropy ................................ ................................ .................. 32 Relevance of Theories of Motor Control to Study 1 ................................ .......... 33 Section 3: Measures to Enhance UE Adaptive Variability Post Stroke ................... 34 Section 4: Effect of Intervention u pon UE Adaptive Variability Post Stroke ............ 35 3 IS REACHING POST STROKE MORE OR LESS VARIABLE?: IT DEPENDS ON HOW ONE DEFINES VARIAB ILITY ................................ ................................ 39 Background ................................ ................................ ................................ ............. 39 Methods ................................ ................................ ................................ .................. 43 Participants ................................ ................................ ................................ ....... 43 Procedures: Set Up and Instrumentation ................................ ......................... 44 Upper extremity (UE) kinematic analysis of reach to grasp ....................... 44 Data processing ................................ ................................ ......................... 46 Kinematic variables ................................ ................................ .................... 46 Standard deviation ................................ ................................ ..................... 47

PAGE 8

8 Surrogate analysis ................................ ................................ ..................... 47 Approximate entropy ................................ ................................ .................. 47 Statistical Analysis ................................ ................................ ............................ 50 Results ................................ ................................ ................................ .................... 51 Discussion ................................ ................................ ................................ .............. 53 Implications for Rehabilitation ................................ ................................ ................. 59 4 DOES MOVEMENT SPEED A ND/OR RHYTHM IMPROVE UPPER EXTREMITY ADAPTIVE VARIABILITY POST STROKE? ................................ ...... 66 Background ................................ ................................ ................................ ............. 66 Methods ................................ ................................ ................................ .................. 71 Participants ................................ ................................ ................................ ....... 71 Procedures: Set Up and Instrumentation ................................ ......................... 71 Upper extremity (UE) kinematic analysis of reach to grasp ....................... 71 Data processing ................................ ................................ ......................... 73 Surrogate analysis ................................ ................................ ..................... 73 Approximate entropy ................................ ................................ .................. 74 Statistical Analysis ................................ ................................ ............................ 76 Results ................................ ................................ ................................ .................... 77 Discussion ................................ ................................ ................................ .............. 79 Implications for Rehabilitation ................................ ................................ ................. 83 5 DOES INTENSE FUNCTION AL TASK TRAINING ENHANCE UPPER EXTREMITY ADAPTIVE VARIABILITY POST STROKE? ................................ ...... 89 Background ................................ ................................ ................................ ............. 89 Methods ................................ ................................ ................................ .................. 92 Research Design for Constraint Induced Motor Treatment Protocol ................ 92 Participants ................................ ................................ ................................ ....... 92 Procedures ................................ ................................ ................................ ....... 9 3 Upper extremity (UE) kinematic analysis of reach to grasp ....................... 93 Data processing ................................ ................................ ......................... 94 Surrogate analysis ................................ ................................ ..................... 95 Approximate entropy ................................ ................................ .................. 95 Constraint induced movement treatment ................................ ................... 98 Statistical Analysis ................................ ................................ ............................ 99 Reliability Analyses ................................ ................................ .......................... 99 Relative reliability ................................ ................................ ....................... 99 Absolute reliability ................................ ................................ ...................... 99 Statistical Testing for Intervention Effects ................................ ....................... 101 Results ................................ ................................ ................................ .................. 102 Testing Determinism in Joint Angles ................................ .............................. 102 Reliability Analyses ................................ ................................ ........................ 103 Relative reliability ................................ ................................ ..................... 103 Absolute reliability ................................ ................................ .................... 103 Standard error of measurement and minimal detectable change ............. 103

PAGE 9

9 Effect of CIMT on UE Adaptive Variability ................................ ...................... 104 Discussion ................................ ................................ ................................ ............ 105 6 CONCLUSION: INTEGRATING THE FINDINGS ................................ ................. 118 LIST OF REFERENCES ................................ ................................ ............................. 123 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 133

PAGE 10

10 LIST OF TABLES Table page 3 1 Participant demographics: individuals with stroke ................................ .............. 61 3 2 Participant de mographics: healthy controls ................................ ........................ 62 3 3 Mean of upper extremity (UE) kin e matic variables ................................ ............. 62 3 4 C orrelations between approximate ent ropy (ApEn) and UE kinema tics ............. 62 4 1 Participant demographics: individuals with stroke ................................ .............. 84 4 2 Participant demographics: healthy controls ................................ ........................ 85 5 1 Participant demographics: reliability analysis ................................ ................... 111 5 2 Participant demographics: intervention analysis ................................ ............... 112 5 3 Reliability estimates ................................ ................................ .......................... 112 5 4 Individual percent change of ApEn in various UE joints after CIMT .................. 113 5 5 Mean change of ApEn in various UE joinits after CIMT ................................ .... 113 5 6 Normalized i ndividual pre and post ApEn in various UE joints ........................ 114 5 7 Fugl Meyer UE subscale and Wolf Motor Function Test scores ....................... 114

PAGE 11

11 LIST OF FIGURES Figure page 2 1 Bell shaped reaching curve ................................ ................................ ................ 37 2 2 Shoulder flexion/extension angle during reach to grasp movement ................... 37 2 3 Elbow flexion/extension angle durin g reach to grasp movement ........................ 38 3 1 Upper extremity (UE) marker set. ................................ ................................ ....... 63 3 2 Standard deviation : healthy and individuals with stroke ................................ ...... 63 3 3 Approximate Entropy (ApEn) healthy and individuals with stroke ....................... 64 3 4 ApEn percent : healthy and individuals with stroke ................................ ............. 65 4 1 UE marker set ................................ ................................ ................................ ..... 85 4 2 ApEn in healthy participants in three reaching conditio ns. ................................ 86 4 3 ApEn in individuals with stroke in three r eaching conditions ............................. 86 4 4 ApEn percent: healthy participants in three reaching conditio ns. ....................... 87 4 5 ApEn percent : individuals with stroke in three reaching conditio ns. ................... 87 4 6 ApEn percent :in grasp task in healthy a nd individuals with stroke ..................... 88 4 7 ApEn : healthy comfortable pace and stroke rhythm conditions. ........................ 88 5 1 Bland an d Altman plot of ApEn shoulder ................................ ......................... 115 5 2 Bland and Altman plot of ApEn elbow ................................ ............................. 115 5 3 Bland and Altman plot of ApEn wri st ................................ ............................... 116 5 4 ApEn of UE joints post Constraint induced movement therapy (CIMT) ............ 116 5 5 WMFT scores post CIMT ................................ ................................ ................. 117 5 6 FM_UE po st CIMT ................................ ................................ ............................ 117

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy UNDERSTANDING VARI A BILITY IN REACHING MOVEMENTS POST STROKE: NON LINEAR DYNAMICAL SYSTEMS PERSPECTIVE By Amit Sethi December 2010 Chair: Lorie G. Richards Major: Rehabilitation Science Stroke is the leading cause of long term disability in the United States and around the world (Duncan, 1995). Upper extremity impairment is one of the most frequent impairments after stroke (Gresham, et al., 1975) The damage to the motor system caused by the stroke results in imperfect motor control, often exhibited as atypical or stereotypical movement patterns. One hallmark of this dysfunctional motor system is the high variability present in several movement parameters, such as upper extremity joint range of motion, movement time, and peak velocity, when variability is conceptualized from the traditional motor control perspective (Cirstea & Levin, 2000) Movement variability under these traditional motor control theories is considered as undesi rable noise in the motor output (Stergiou, Buzzi, Kurz & Heidel, 2004), and therefore error. However, contemporary motor control theories, such as dynamic systems theory, consider variability as a n intrinsic characteristic of movement and plays an integral role in motor learning ( Bernstein, 1967 ; Kamm, Thelen, & Jensen, 1990) Developing variability is indicative of the development of greater functionality in the motor system. Variability in healthy biological systems reveals the inherent complexity of the system components and their functional interactions (Vaillancourt & Newell, 2002) Further,

PAGE 13

13 variability in various physiological systems not only explains the complexity of a healthy system but also reflects adaptability and flexibility to the system. It might then seem optimal complexity as a function of aging and disease. Because of the damage to motor neural networks from stroke, it would then seem intuitive that individuals with stroke might also exhibit reduced complexi ty or adaptive variability in upper extremity movement. If this assumption holds true, enhancing upper extremity adaptive variability might seem to be an important goal of upper extremity stroke rehabilitation. Therefore, the overall purpose of this disser tation is to examine the adaptive variability of movement in the upper extremity and understand its relationship to upper extremity motor performance post stroke. This dissertation comprises three studies. The first study investigates whether the adaptive variability of upper extremity joint kinematics is reduced post stroke. The second study examines some of the task variables that might augment upper extremity adaptive variability post stroke. The final study investigates the change of upper extremity ada ptive variability in individuals with stroke following an intervention based upon Constraint Induced Movement Therapy.

PAGE 14

14 CHAPTER 1 INTRODUCTION Stroke is the leading cause of long term disability in the United States and around the world (Duncan, 1995). The American Heart Association (2010) reports over 795,000 new cases of stroke annually in the United States. Stroke is estimated to result in $30 billion in health care costs and lost productivity each year and the incidence of stroke is not decreasing (Han & Haley, 1999) With the aging population, this suggests that the numbers of individuals disabled by s troke will only rise in the future, increasing societal burden. Upper extremity (UE) impairment is one of the most frequent impairments after stroke (Gresham, et al., 1975) Thirty to sixty per cent of individuals with stroke are unable to use their more affected upper extremities in functional activities (Gresham, et al., 1975) Despite the billions of dollars spent on UE rehabilitati on each year, residual UE impairments still exist (Duncan, Lai, & Keighley, 2000; Nakayama, Jorgensen, Raaschou, & Olsen, 1994) Therefore, it is imperative to discover more effective UE rehabilitation intervention strategies. A more thorough understanding of the post stroke motor impairments is needed to develop more effective treatments that will maximize motor ability and enhance functional independence. Examining the impaired motor system provides a window to further comprehend the UE motor control deficits following stroke. UE motor impairments post stroke often include impaired motor control and reduced inter joint coordination (Cirstea & Levin, 2000) Among the umbre lla of motor impairments, individuals with post stroke hemiparesis often exhibit atypical or stereotypical movement patterns. The resulting movements exhibit high variability in several movement

PAGE 15

15 parameters, such as UE joint range of motion, movement time, and peak velocity, when variability is conceptualized from the traditional motor control perspective (Cirstea & Levin, 2000) Movement variability under these traditional motor control theories is considered as und esirable noise in the motor output (Stergiou, Buzzi, Kurz & Heidel, 2004), and therefore error. Hence, conventionally, a major goal of rehabilitation interventions is to reduce movement variability, with the intent to approximate normal movement (Bobath, 1 990). However, variability is often intrinsic to the outcome of movement (Newell, Deutsch, Sosnoff & Mayer Kress, 2005). Movement variability is also associated with the motor behavior of human beings and is a consequential element of our motor control. C ontemporary motor control theories, such as dynamical systems theory, consider variability as a positive characteristic of movement and plays an integral role in motor learning (Bernstein, 1967; (Kamm, et al., 1990) Optimal variability reflects the greater functionality of the motor system. Hence, understanding movement variability might be a mechanism to better comprehend the motor system. Further, variability in healthy biological systems reveals the inherent complexity of the system components and their functional interactions (Vaillancourt & Newell, 2002) For example: the force output of a hand muscle might depend upon several variables: n umber of motor units recruited, gripping surface, level of motivation, etc. Thus, force production is dependent upon the complex interplay of these components, the coordination of which again reflects the health or functionality of the motor control system As a healthy system is a complex system, variability could then be attributed as

PAGE 16

16 a descriptor of complexity, and therefore a measure of the optimal functioning of a system. Complexity is also essentially integrated in healthy physiological systems of ou r body (Glass & Mackey, 1988; Goldberger, Rigney, & West, 1990) For instance, greater variability in the heart and brain waves is correlated with a healthy state (Elbert, et al., 1994) Greater complexity in heart rate suggests the moment to moment fluctuations of our hearts during the day (Lipsitz & Goldberger, 1992; Pool, 1989) These fluctuations are e ssential to adapt to the diurnal changes in heart rate with physical activity. Similar changes in heart rate are also observed with increase in physical activity such as, running or biking. Healthy individuals without any cardiac anomalies are able to adap t to the increased physiological demand caused due to the change in physical activity. A healthy cardiovascular system demonstrates an optimal level of complexity characterized by dynamic, constantly adapting or less regular heart activity as measured by t he electrocardiogram (Goldberger, et al., 1990; Lipsitz & Goldberger, 1992) Likewise, complexity of an intact motor system depicts the inherent abundance of the options to perform a motor task (Latash & Anson, 2006) The greater the number of options, the greater the ability to adapt to changing environmental demands and the greater the ability to emit an appropriate adaptive response to those demands. For instance, one could employ multiple strategies t o pick up a glass from the surface of the table. Complexity of movement provides individuals with an abundant repertoire of movement strategies to adapt movement patterns and successfully meet the demands of everyday changing tasks ( Newell, et al., 2005; H arbourne & Stergiou, 2009) Similar to the cardiovascular system, an intact motor system demonstrates dynamic, and less

PAGE 17

17 regular motor output (such as force generated by muscles) (Sosnoff & Newell, 2006b) Thus, complexity in healthy physiological system s is characteristic of an adaptable and flexible system. It might then seem intuitive to refer the complexity of the syst em as, Apart from mo tor control and behavior, complexity or adaptive variability is also considered consequential in multiple physiological systems of the human body ( Glass & Mackey, 1988; Goldberger, 1986) complexity hypot and disease. In general, aging and pathology are characterized by remarkable regular or less dynamic physiological and motor responses (Goldberger, 1997) For instance, older individuals and individuals with cardiac disorders show a regular pattern of heart rate exhibiting reduced complexity (Goldberger, 1997). Likewise, resting tremors observed in Parkinsonism also exhibit reduced compl exity secondary to rigid and stereotypical movement pattern s (van Emmerik, Sprague, & Newell, 1993) UE kinematic studies also reveal st ereotypical movement patterns in individuals with stroke (Cirstea & Levin, 2000) Although these movement patterns have not been examined for complexity, according to the loss of complexity hypothesis it would then seem intuitive that individuals with stroke might also demonstrate reduce d complexity or adaptive variability in UE movement. Understanding the adaptive variability in UE movements could offer a unique approach to examine the health or functionality of the motor control system post stroke and offer additional ways to describe t he impairments in motor control post stroke. If individuals with stroke exhibit reduced UE adaptive variability, enhancing UE adaptive variability might constitute one of the important goals of UE

PAGE 18

18 stroke rehabilitation. UE adaptive variability might restor e the ability to adapt movement patterns and successfully make adjustments to the dynamic nature of functional tasks in individuals with stroke Although not investigated, existing UE rehabilitation interventions might augment UE adaptive variability post stroke. One could also question if g ains made in UE adaptive variability translate to improvements in function and participation post stroke The overal l purpose of this dissertation wa s to examine the adaptive variability of UE movement and understand its relationship to UE motor performance post stroke. This dissertation is comprised of three studi es. The first study investigated whether the adaptive vari ability of UE joint kinematics wa s reduced post s troke. The second study examined some of the task par ameters that might augment UE adaptive variability post stro ke. The final study investigated the change in UE adaptive variability in individuals with stroke following an intervention based upon Constraint Induced Movement Therapy (Wolf et al., 2006) The research proposed is signific ant because it not only provide s a novel understanding of the motor impairment by characterizing UE adaptive variability, b ut also examine s whether a commonly researched UE intervention augments UE adaptive variability post str oke.

PAGE 19

19 CHAPTER 2 THEORETICAL BACKGROU ND AND SIGNIFICANCE This chapter provides the theoretical background and underlying specific aims of the proposed research. This chapter is divided into four sections. The first section describes upper extremity ( UE ) mo tor impairments and the current understanding of movement variability that results from damage to the motor system post stroke. The second section reviews the theoretical background of variability in motor control and also briefly describes the measures fo r quantifying movement variability. This section also provides the specific aims for Study One (Chapter 3). The third section discusses potential ways of augmenting variability post stroke and forms the basis of aims for Study Two (Chapter 4). The last sec tion discusses the effect of intervention in augmenting variability post stroke and also provides aims for Study Three (Chapter 5). Section 1: Upper Extremity (UE) Impairment and Movement Variability in S troke Stroke: Incidence and Consequences dysfunction due to an abnormality in cerebral circulation with resultant signs and new strokes occur an nually in the United States (American Heart Association, 2010). S troke is the leading cause of serious long term adult onset disability in the United States (American Heart Association, 2010) and the second worldwide in individuals more than 60 years of ag e (Barnes, Dobkin, & Bogousslavsky, 2005). The damage to the motor system caused by stroke most often results in motor impairments and UE impairment is one of the most frequent impairments after stroke (Barnes, Dobkin, & Bogousslavsky, 2005). The impairmen t of the more affected UE causes difficulty with

PAGE 20

20 daily activity completion and decreases quality of life (Gresham, et al., 1975) The incidence of stroke is not decreasing. With the aging popula tion, the human and financial burden of stroke will continue to increase over time. Pathophysiological Basis of UE Impairment and Performance Post Stroke The pathophysiological basis of UE motor impairment and performance post stroke occurs due to death of neurons and the disruption of the motor neural networks and pathways of the central nervous system caused by interruption of arterial blood supply from a hemorrhage (hemorrhagic stroke) or clot (ischemic stroke) usually on one side of the brain (Barnes, D obkin, & Bogousslavsky, 2005). The resulting impairment leads to paresis (or paralysis) in the opposite half of the body (hemiparesis) (Barnes, Dobkin, & Bogousslavsky, 2005). The types and degrees of disability that follow a stroke depend primarily upon m ultiple factors including: location and size of brain lesion, severity of the lesion, individual degree of spontaneous recovery, and the duration of stroke onset (Alexander, 1994; Thirumala, Hier, & Patel, 2002) R esidual deficits are common after a stroke and might include: sensorimotor, cognitive and visual deficits, all of which can independently or in combination result in reduced or impaired UE motor performance. Motor control impairments of weakness (paresis), loss of volitional movements of the weaker or paretic side (opposite to lesion) or inappropriately graded muscle activations of the weaker side affect UE motor performance immediately after a stroke (Barnes, Dobkin, & Bogousslavsky, 2005). Spasticity/hype rreflexia and changes in the mechanical properties of muscles further contribute to impaired UE motor performance, developing a few weeks after the initial insult (Sahrmann & Norton, 1977)

PAGE 21

21 Motor Recovery P ost S tr oke and Movement V ariability M ovement patterns exhibited during motor recovery are often stereotypical and depict high variability in UE movement parameters when variability is conceptualized from the traditional motor control perspective. For instance, hi gh variability was observed in UE joint range of motion, movement time, and peak velocity post stroke (Cirstea & Levin, 2000) Furthermore, variability of wrist and finger extension force was also observed to be greater post stroke as compared to individuals with no neurological impairments (Lodha, Naik, Co ombes, & Cauraugh 2010 ) In this context, greater variability indicates greater error of the respective movement parameter (kinematics, kinetics or muscle activation) and compr om ised motor control (Lai, Mayer Kres s, & Newell, 2006; Slifkin & Newell, 1999) Hence, conventionally, a major goal of rehabilitation interventions is to reduce movement variability, with the intent to approximate normal movement (Bobath, 1990). The next section discusses the theoretical un derpinnings of movement variability. Section 2: Theoretical B asis of M ovement V ariability Variability in Movement Past a nd Current Perspective Movement variability has been a topic of debate in human movement sciences, where dichotomous opinions exist. Traditional theories and models of motor control consider variability as error whereas contemporary perspective acknowledges the functional role of variability in motor development and lea rning. This section provides an overview of both traditional and co ntemporary view of motor control and learning. Traditional view of motor control and learning The traditional theories of motor control and learning did not acknowledge the functional role of variability in human movement science (Newell & Slikfin, 1998) The

PAGE 22

22 traditional perspective of motor control was influenced by the information processing theory of communication systems (Shannon & Weaver, 1949). According to the information processing theory, a signal is distorted due to addition of the noise during transmission such that the observed motor output is a combinatio n of signal and noise Based on this tenet, movement variability associated with neurological disorders such as brain injury and stroke is also considered as error. This basic principle of inf ormation processing theory influenced human movement sciences and led to the development of closed and open loop model s of motor control (Schmidt & Lee, 2004). The c losed loop model of motor control views the human information processing in a top down app roach, where the executive regions of the body (brain) regulate the movement mechanism (Adams, 1971) The reference mechanism then compares this input to the information obtained from the environment (feedback) after completion of movement, and computes an error, representing the difference between the actual and desired states. The error thus, computed then provides information to the executive level (brain), where decisions are made about how to reduce the error. Thereafter, the executive level provides instructions to the effector level (extremities) and results in movement in an environment. Further, the errors associated with the resulting movement are aga in fed into the reference mechanism via feedback obtained from the environment. Hence variability in the desired motor output is considered as error and However, slow processing via feedback in the closed loop model cannot explain rapid movements (Schmidt & Lee, 2004). As a result, rapid and ballistic movements

PAGE 23

23 were explained by an open loop control model or motor program (Keele, 1968) Open loop control suggests that, a pre planned set of instructions, or motor program, exists to execute rapid movements without involving feedback during the movement. Feedback is only used post movement to compa re intended outcomes with actual outcomes for use in programming the subsequent movements. Again, the idea was to decrease problems of motor programs for different moveme nts. Specifically, movement scientists argued that the human brain could not produce novel motor programs and further store them for multiple movements. I (Schmidt, 1975) proposed the S chema T heory of discrete motor learning. Schema theory proposes a generalized motor program (GMP) that is an abstract representation comprising the general characteristics for a particular class of actions. The GMP is stored in memory and results in a unique pattern of activity whenever the program is executed (Schmidt, 2003). For instance, Lashley (19 42) and Bernstein (1967) asked individuals to write dominant hand; feet; and teeth. Although the writing samples differed in the amount of accuracy the spatial pattern remained similar or in variant across various effectors (Bernstein, 1967). These findings suggest that certain parameters of movement remain invariant and are rigidly structured in the GMP (Schmidt, 2003) In particu lar, the order of events, relative timing and relative force of movement are believed to be the invariant parameters of GMP (Schmidt, 2003) Therefore, based on the invariant characteristics of GMP, variance in the motor output is attributed to changes in the parameters of GMP resulting in an

PAGE 24

24 erroneous movement. This assumption further strengthens the notion that variability in motor output is representative of error. The r ole of variability h as also been criticized in motor learning. Fitts and Peterson (1964) also proposed three stages of motor learning. According to this theory, motor learning can be viewed as a continuous process with gradual changes in the nature of information processing a s learning progresses (Davids, Button & Benett, 2008). In particular, an individual learns novel skill acquisition following three stages of motor learning: cognitive stage, associative stage and autonomous stage. During the initial cognitive stage, an ind ividual is exposed to simple rules and verbal instructions to acquire basic understanding of the novel task. As the individual tries multiple configurations to accomplish the task, the is characterized with hig h variability resulting in large errors. Further practice of the task, reduces the errors in movement and results in consistent and refined movement patterns in the associative stage of motor learning. Lastly, the autonomous stage is characterized with ext ensive practice, further resulting in reduced errors requiring minimal mental effort to produce the movement (Schmidt & Lee, 2004). Fitts & (1964) theory of motor learning theory also parallels the GMP where, extensive practice and experience re sults in the refinement of GMP, and reduction in errors further results in the development of schema for specific movement (Schmidt, 2003) In general, traditional theories and models of motor control and learning viewed variability in movement as error. Additionally, an individual progressed through the stages of motor learning by reducing variability (error) in movement. Traditional motor

PAGE 25

25 control and learning theories primarily explained the movement of discrete motor tasks (Schmidt, & Lee, 2004). Therefore, traditional theories employed linear models in understanding variability. Linear models of variability quantify the amount or magnitude of variability alone (Newell, et al., 2005). Specifi cally, these models provide information about the amount of error associated with movement. Linear measures of variability Two commonly employed measures based on l inear models are : standard deviation, and coefficient of variation (Newell, et al., 2005). B oth these measures quantify variability with respect to the mean of a sample. These measures are described in the following section. Variance/ s tandard deviation (SD) Variance and SD measure the consistency in the movement outcome (Schmidt & Lee, 2004). Fu rther variance is a measure of variability that employs the sum of the squared deviations between the individual values and the sample mean divided by the appropriate degrees of freedom for the sample (Stergiou, et al., 2004). Variance can be computed as: s 2 = n i=1 (x i M) 2 /n 1 where, n is the total number of trials, x i is the value of the independent variable at that trial and M is the mean of the values of the independent variable of all the trials. Standard deviation is the square root of variance, gi ven by SD = [(s 2 )] 1/2 score on each trial and his or her own mean score (Stergiou, et al., 2004). Therefore, a consistent performance on each trial results in small SD. Very consistent performan ce resulting in the same score in each trial results in zero SD. SD thus is an indicator of the error or inconsistency in the movement. SD can also be used to construct the

PAGE 26

26 confidence intervals across the mean of an independent variable (Stergiou, et al., 2004). Coefficient of variation (CV) CV is defined as the amount of normalized variability relative to the magnitude of the mean (Stergiou, et al., 2004). CV is computed as: CV = (SD/M) 100, where SD is the standard deviation and M is the sample mean. C V thus results in a dimensionless number, which is useful in the comparison of data sets with different units or different means (Stergiou, et al., 2004). For example, m ovement characteristics of healthy seen with pathology such as stroke are random or erroneous. Both SD and CV have been used to study the deviations in the patterns of movement in individuals with stroke. For instance, Woodbury, et al.( 2009 ) compare d individuals with stroke against healthy controls during UE functional movements. The SD of multiple kinematic measures of UE kinematics such as : number of velocity peaks, shoulder and elbow joint excursions, trunk displacement and index of curvature reve aled significantly greater amount of variability in individuals with stroke than healthy controls. SD and CV of wrist and finger extension force output were also observed to be greater in individuals with stroke than healthy controls (Lodha, et al. 2010 ) Based upon the principles of traditional motor control theories and linear models, the conventional rehabilitation approach in individuals with stroke has been to curtail the past two dec ades the field of motor control and learning has developed another school of thought, which considers movement variability essential to skill acquisition and skilled learning (Newell, 1986).

PAGE 27

27 Contemporary view of motor control and learning During the last decade movement science research based on dynamical systems theory (DST) has proposed an alternative perspective to the understanding of motor control and learning. DST believes variability as essential and plays a functional role in the process of motor l earning (Davids, Button & Benett, 2008; Kamm, et al., 1990; Scholz, 1990 ) Unlike the traditional perspective, DST is less dependent upon the assumption of sequential skill acquisition related to the hierarchical m aturation of neural structure. DST views movement systems as complex where movement pattern emerges from the dynamics of the interactions among the multiple components of the system utilizing inputs from both higher brain center s and the environment (David s et al., 2008; Kelso, 1981, 1984, 1995; Scholz, 1990). Further, following a change in the environmental conditions the movement system undergoes a transient state of disequilibrium, and ultimately the underlying components of the system may self organize into another optimal movement pattern (Haken, Kelso, & Bunz, 1985) For example, healthy individuals maintain a stable gait pattern while walking on a non slippery surface. In contrast, walking on an icy surface mi ght result in initial loss of balance or instability of gait. However, the gait pattern again becomes stable gradually after making several adjustments on the icy surface. The initial loss of stability might be accompanied by increased variability in the o verall gait when variability is conceptualized from the traditional motor control perspective However, variability reduces with learning as the overall gait pattern becomes more stable. Thus, a change in environmental condition alters the movement variabi lity while transitioning the system from one stable state to another state of stability and plays a consequential role in motor learning (Bernstein, 1967).

PAGE 28

28 Bernstein (1967) also raised an empirical question of how the brain controls multi joint movements. Bernstein (1967) introduced the concept of degrees of freedom, which are independent parts of the moving body that must be organized to achieve a task goal. For instance, driving requires an individual to coordinate both hands and foot to steer a car whil e braking or moving the wheel (Davids, et al., 2008). Specifically, process of mastering redundant degrees of freedom of the moving organ, in other words its conversion to movement utilizing multiple degrees of freedom suggesting the inherent motor redundancy in humans (Bernstein, 1967). Based upon this contemporary perspective motor learning occurs in thre e stages (Bernstein, 1967). The first stage involves freezing of degrees of freedom. During this stage, initial practice of a novel motor task limits the degrees of freedom of the limbs while reducing the overall movement variability. The next stage of mot or learning is associated with releasing and reorganizing degrees of freedom. Specifically, with practice, the constraints on the degrees of freedom are reduced, allowing independence in movement and higher level of success. The final stage includes exploi ting the mechanical inertial properties of the limbs. In particular, in this stage the learner takes advantage of the intrinsic mechanical inertial properties of the limbs in performing highly skilled movements. The third stage is particularly evident in a thletes, dancers, jugglers and other individuals performing skilled motor tasks (Davids, et al., 2008). The skilled movements performed by these individuals demonstrate increased variability in the multiple components of the movement while maintaining an o ptimal level of control at

PAGE 29

29 the desired outcome of the movement (Davids, et al., 2008). Vereijken, Van Emmerik, Whiting and Newell (1992) demonstrated this phenomenon of change in degrees of freedom in seven individuals practicing slalom like ski movements on a ski apparatus for 7 consecutive days. Early practice of the task resulted in reduced angular excursions of the hip, knee, ankle and torso. Further, high cross correlations were also observed between the ipsilateral joints of the lower extremities sugg esting high intra limb couplings resulting in freezing degrees of freedom. However, during the late stages of practice the joint angular excursions of hip, knee, ankle and torso increased significantly. In addition, a decrease in the cross correlation betw een the ipsilateral joints of the lower extremities suggested release of degrees of freedom. This classic experiment illustrated the functional role of variability in learning a novel motor task. Apart from motor control and behavior, variability is also c onsidered consequential in multiple physiological systems of the human body ( Glass & Mackey, 1988; Goldberger, 1986) For instance, greater variability in the heart and brain waves is correlated with a healthy state (Elbert, et al., 1994; Lipsitz & Goldberger, 1992) Greater heart rate variability suggests the moment to moment fluctuations of our hearts during the day (Pool, 1989) These fluctuations are essential to adapt to the diurnal changes in heart rate. Similar changes in heart rate are also observed with increase in physical activity such as, running or biking. Healthy individuals without any c ardiac anomalies are able to adapt to the increased physiological demand caused due to the change in physical activity. Lipsitz and Goldberger, ( 1992) observed differences in heart rate variability between a young and elderly individual. Specifically, t he m oment to moment fluctuations wer e more evident in the heart wave of a young individual indicating greater

PAGE 30

30 heart rate variability. A healthy cardiovascular system demonstrates an optimal level of complexity characterized by dynamic, constantly adapting or l ess regular heart activity as measured by the electrocardiogram (Lipsitz & Goldberger, 1992; Pool, 1989) Variability in various physiological systems is associated with the complexity of a healthy system and is ch aracteristic of an adaptable and flexible system. Variability could then be considered to reflect the complexity of the behavioral and physiological system. Complexity can be quantified using non linear models. Non linear models are not the determinant of the magnitude of variability and thus, do not consider variability as error. Non linear models measure the structure of variability (Newell, et al., 2005). The structure of variability is the t emporal organization of a movement pattern where values emerge in an orderly manner over a period of time (Harbourne & Stergiou, 2009). While linear models provide a discrete measure of error associated with the deviation from the mean of the movement (i.e. kinematics, kinetics or EMG) signal, non linear models detec t variability at each time point of the entire continuum of the movement signal. Understanding complexity of movement Stergiou, Harbourne and Cavanaugh (2006) proposed that an optimal variability or complexity is associated with a healthy motor system Op timal complexity is characterized by highly compl ex, yet organized dynamic structure Optimal complexity provides healthy individuals a rich repertoire of motor behaviors and strategies to accomplish tasks. For instance: we employ different type of contrac tions of muscles and range of motion to pick an object from a level surface than for overhead reaching of the same object. In addition, optimal complexity also imparts adaptability to modify movement in the event of perturbations. For instance: we modify o ur grasp patterns

PAGE 31

31 depending upon the size and shape of the objects. Furthermore, while holding a wet glass filled with water, our grasp becomes stronger as soon as we realize that the object might slip in our hands to prevent a spill. Hence, optimal comple xity not only offers multiple options for task performance but also provides adaptability in movement. A decline or loss of this optimal complexity results in a more rigid or stereotyped motor system. A rigid motor system is characterized by regular move ment pattern with ve ry limited variability In contrast, an increase beyond optimal complexity results in a noisy and unstabl e system Complexity at both extremes makes the system less adaptable to perturbations (Ha rbourne & Stergiou, 2009; Stergiou, Harbourne, & Cavanaugh, 2006) Aging and disease are associated with a loss of physiological and behavioral complexity resulting in maladaptive responses to everyday stresses and perturbations (Lipsitz & Goldberger, 1992) For instance, resting tremors observed in Parkinsonism exhibit regular or stereotypical pattern and reveal less physiologic complexity (van Emmerik, et al., 1993) Vailliancourt and Newell (2002) suggested that change (increase/decrease) in optimal complexity is task dependent. They postulated that lower complexity is considered optimal in rhythmic tasks such as walking. Conversely, greater complexity is optimal in fixed point tasks like reaching. Despite the task, healthy motor behavior is marked by optimal complexity which confers adaptability and thus complexity could be referred as, Adaptive variability and stroke The l oss of complexity hypothesis suggests a decline in optimal complexity as a function of aging and pathology (Lipsitz & Goldberger, 1992). According to the loss of complexity hypothesis it would t hen seem intuitive that individuals with stroke might also

PAGE 32

32 depict reduced adaptive variability in UE movement. Further, diminished UE adaptive variability might limit the adaptive capacity of the motor system to self organize into optimal movement patterns following perturbations post stroke. For instance, individuals with stroke might not be able to modify grasp patterns to accommodate a change in size and shape of the objects. Understanding complexity in UE movement s using non linear tools Advanced mathe matical techniques inspired by chaos theory provide non linear measures to examine the complexity in motor behavior. Application of non linear analyses could also be applied to discreet UE movements. Morasso ( 1981 ) demonstrated that the velocity profile of the hand during reaching is a bell shaped curve ( f igure 2 1 ). Figures 2 2 and 2 3 also show the shoulder and elbow joint angles of healthy individuals while performing a reach to grasp movement. These examples suggest that UE kinematics exhibit non line ar characteristics and underscore the appropriate use of non linear techniques in understanding the complexity in UE kinematics. Approximate entropy Approximate entropy (ApEn) is one of the non linear techniques used to quantify complexity (Lipsitz & Goldberger, 1992) The most common method employed in the computation of ApEn is to identify repeating patterns of length ( m) across short lengths of the time series of interest (i.e. joint angle) Starting wi th a vector of length m at point pi in the time seri es the procedure involves counting the number of other vectors at within r times the standard deviation of the time series. As a result, Cm(r) a count of the recurrence of vectors of length m is obtained. Thereafter, the same procedure is carried out by

PAGE 33

33 repeating patterns iteratively evaluating time series segments of length (m) and the logarithm of the results is summed r is a similarity criterion, and provides the limit s for asses sing the nearness of adjacent data points in the time series Another parameter, lag, identifies the number of time skips between points in one of the length m vectors. Biomechanical data analysis conventionally utilizes lag = 1, r = 0.2 times the standard deviation of the time series, and m = 2 (Slifkin & Newell, 1999) Thereafter, the log of this similarity count Cm(r) is normalized by the number of points in the time series. This quantity is followed by the recur rence of vectors of length m +1 Cm+1(r ), in the entire time series. ApEn is thus computed as the natural logarithm of the ratio of Cm(r) and Cm+1(r), as follows: ApEn (X m.r) = log [C m (r)/C m+1 (r)] (2 1) In general, a vector of shorter length repea ts more often than a longer one within a time series, thus the lowest possible ApEn value can be the natural logarithm of 1, which is 0, and negative values might not be obtained. In a highly periodic or regular time series, values of Cm(r) can be similar to Cm+1(r) Hence, smaller values characterize a more regular time series where similar patterns are more likely to follow one another suggesting low complexity (adaptive variability). In contrast, high ApEn values, suggest a highly irregular time series, where the predictability of subsequent patterns is low, suggesting high complexity (adaptive variability). Relevance of Theories of Motor Control to Study 1 Two varying viewpoints exist about movement variability: the traditional linear perspective and the contemporary dynamical non linear perspective. In fact the two opposing viewpoints are not conflicting, but actual ly complementary to each other. These two viewpoints actually measure two distinct dimensions of variability. The

PAGE 34

34 traditional viewpoint prima rily quantifies the amount of error associated with the movement or magnitude of variability (Newell et al., 2005). The contemporary viewpoint, on the other hand, measures the structure of variability. The structure of variability reveals the temporal orga nization of a movement pattern where values emerge in an orderly manner over a period of time (Harbourne & Stergiou, 2009) Further structure of variability reveals the inherent complexity of the system components a nd their functional interactions (Vaillancourt & Newell, 2002) Movements after central nervous system damage or dysfunction, for instance in movement disorders such as tardive dyskinesia, show decreased com plexity (van Emmerik, et al., 1993; Vanemmerik, Sprague, & Newell, 1993) While the magnitude of variability in UE kinematics post stroke has been studied (Cirstea & Levin, 2000; Woodbury, et al., 2009) complexity has not been explored. Because from a contemporary perspective measures of variability offer windows into the functionality of the motor control system, a complete understanding of the impact of stroke on variabili ty and complexity and its relationship to UE function may serve to foster development of better interventions for the damaged motor system. Stud y One of this dissertation aimed to examine whether the complexity or adaptive variability as measured by ApEn i n shoulder, elbow, wrist and proximal interphalangeal (f lexion/extension) joint angles wa s decreased post stroke. Further, the relationship between adaptive variability and UE motor performance was also explored. The knowledge gained from this study would provide information about the adaptive capacity of UE post stroke Section 3: Measures t o Enhance UE Adaptive Variability Post Stroke After exploring the deficits in UE adaptive variability post stroke, the next logical step is to identify variables to aug ment this adaptive capacity. D ecreased UE adaptive

PAGE 35

35 variability might result in s tereotypical movement patterns post stroke. According to Dynamical systems theory (DST), these stereotypical patterns could be referred as (Kamm, et al., 1990) Consistent with the notion of attractor states, individuals with stroke fall into a stereotyped pattern easily and return to that pattern even when perturbed or interrupted (Kamm, et al., 1990) Utilizing the principles of DST, UE adaptive variability might be enhanced by changing certain task constraints, known as control parameters (Newell, 1986). Task constrain ts or control parameters are variables that might be highly specific, such as myelination, particular muscle strength, movement speed or nonspecific, such as emotional or motivational aspects. Changing the appropriate task constraints might transition the stable or stereotypical movement patterns with low adaptive variability to more variable and adaptable ones. Movement speed has been shown to drive the system from one stable state (or movement pattern) to another in both bimanual reaching and walking (Diedrich & Warren, 1995; Kelso, 1984) In addition, reaching with rhythmic cues also showed immediate improvement in movement composition and UE kinematics post stroke (Thaut, Kenyon, Hurt, McIntosh, & Hoemberg, 2002) However, the immediate effects of movement speed and auditory rhythmic cuing as a mechanism to enhance the adaptive variability of UE post stroke have not been studied. Therefore, Study Two of this dissertation aimed to investigate whether the ada ptive variability of UE, as measured by ApEn, was enhanced immediately when reaching naturally at a faster than normal speed and/or to auditory rhythmic cues in individuals with stroke Section 4: Effect of Intervention upon UE Adaptive Variability Post St roke Low UE adaptive variability might affect the functional use of the UE because an optimal amount of adaptive variability provides individuals with an abundant repertoire

PAGE 36

36 of movement strategies to adapt movement patterns and successfully meet the demand s of everyday changing tasks. Stergiou, Harbourne and Cavanaugh (2006) indicated that one of the goals of neurological rehabilitation should be to facilitate the adaptive variability. Among the common repertoire of UE rehabilitation interventions post stro ke, Constraint I nduced M ovement T herapy (CIMT) has been studied most widely (Langhorne, Coupar, & Pollock, 2009) CIMT is a combinati on of various intervention principles aimed to enhance the use of the more affected UE in individuals post stroke. Systematic reviews suggest the highest level of evidence of CIMT in UE stroke rehabilitation (Bonaiu ti, Rebasti, & Sioli, 2007; Langhorne, et al., 2009) A few studies have also shown kinematic changes post CIMT (Caimmi, et al., 2008; Wu, Lin, Chen, Chen, & Hong, 2007) For instance, (Wu, et al., 2007) demonstrated positive changes for UE kinematic characteristics in individuals with stroke post CIMT. Specifically, post CIMT, participants showed significant improvements in reaction time and produced smoother and straighte r movement trajectories with the more affected UE as compared to those who received conventional therapy. Caimmi, et al., (2008) also observed favorable changes in affected UE kinematics in individuals with stroke post CIMT. However, the effect of intense functional task practice based upon CIMT on UE adaptive variability using ApEn has not been studied. Therefore the third study of this dissertation examined the effect of CIMT upon UE adaptive variability post stroke.

PAGE 37

37 Figure 2 1 Bell shaped reaching curve Figure 2 2 Shoulder flexion/extension angle during reach to grasp movement

PAGE 38

38 Figure 2 3 Elbow flexion/extension angle during reach to grasp movement

PAGE 39

39 CHAPTER 3 IS REACHING POST STR OKE MORE OR LESS VAR IABLE?: IT DEPENDS O N HOW ONE DEFINES VARIAB ILITY B ackground Stroke is a leading cause of disability in United States, with over 795,000 individuals affected by stroke every year (American, Heart Association, 2010). Up to 85% of individuals with stroke exhibit hemiparesis, resulting in upper extre mity (UE) impairments immediately post stroke (Olsen, 1990) Despite rehabilitation, residual UE impairments still exist (Duncan, et al., 2000; Nakayama, et al., 1994) A more thorough understanding of the post stroke motor system and the associated impairments is needed to develop more effective treatments that will maximize motor ability post stroke and enhance functional independence. Among the umbrella of motor impairments, individuals with post stroke hemiparesis often exhibit atypical movement patterns. The resultin g movements exhibit high variability in several movement parameters, such as UE joint range of motion, movement time, and peak velocity, when variability is conceptualized from the traditional motor control perspective (Cirstea & Levin, 2000) Movement variability under these traditional mo tor control theories is considered as undesirable noise in the motor output (Stergiou, et al., 2004), and therefore error. Hence, conventionally, a major goal of rehabilitation interventions is to facilitate recovery of the motor system and, therefore, red uce movement variability, with intent to approximate normal movement (Bobath, 1990). In contrast, contemporary motor control theories, such as dynamical systems theory (DST), consider movement variability as a positive characteristic of a healthy motor sy stem and plays an integral role in motor learning (Kamm, et al., 1990) For

PAGE 40

40 instance, Vereijken and colleagues (1992) demonstrated that during the early stage of practicing of a novel task, such as performing slalom like movements on a ski simulator, the variability of extremity joint angles is reduced (Vereijken, Vanemmerik, Whiting, & Newell, 1992) However, after many hours of practice the variability of joint angles increase d allowing independence in movement across joints and higher levels of success. DST purports that greater variability provides individuals with repertoire of movement strategies to accomplish a task (Newell, et al., 2005). Greater variability allows movemen t to be adaptable such that individuals have the ability to tolerate perturbations and adjust movement patterns and successfully meet the demands of everyday changing tasks. For example, reaching for a soda can on a level surface would employ different mo vement strategies than overhead reaching for the same task. Hence, understanding movement variability might be a mechanism to better comprehend the motor system. Stroke impairs the ability to perform such fluid and adaptable UE movements and tends to be c haracterized by stereotypical patterns (Cirstea & Levin, 2000) From the DST perspective, one would hypothesize that post stroke movement patterns have low variability. If true, it would then seem intuitive that a goal for therapy should be to increase movement variability in individuals wi th stroke to enhance UE motor control and function. Although it seems as if two contrasting viewpoints exist about movement variability, the two opposing viewpoints are, in fact, not conflicting, but actual ly complementary to each other. These two viewpoi nts actually measure two distinct dimensions of variability. The traditional viewpoint primarily explains the movement of

PAGE 41

41 discrete motor tasks (Schmidt & Lee, 2004) and employs linear analyses in understanding variability. It also quantifies the amount of error associated with the movement or magnitude of variability (Newell et al. 2005). Specifically, variability is measured by computing the standard deviation (SD) or coefficient of variation (CV) of the movement variable, which is calculated around the me an of the entire range of values of that variable (Newell, 1976) Several studies indicate that UE biomechanical movement parameters such as, UE joint range of motion, movement time, and peak velocity post stroke demonstrate greater magnitude of variability as compared to healthy controls (Cirstea & Levin 2000; Woodbury, et al., 2009) O n the other hand t he contemporary viewpoint, characterizes the structure of variability. Structure of variability is the t emporal organization of a movement pattern where values emerge in an orderly manner over a period of time (Harbourne & Stergiou, 2009) Structure of variability utilizes the entire range of values of a movement parameter as opposed to magnitude. For instance, structure of variability could be measured by examini ng the shoulder joint angle over several trials and is thus different than SD and/or CV, which only measure the deviation from the mean of trials. Further structure of variability reveals the inherent complexity of the system components and their function al interactions (Vaillancourt & Newell, 2002) For example: the force output of a hand muscle might depend upon several variables: number of motor units recruited, gripping surface, level of motivation, etc. As a result, force production is dependent upon the complex interplay of these components, Approximate entropy (ApEn) has been commonly utilized to quantify complexity inherent in the motor system (Sosnoff & Newell, 2006a) The information gained from analysis of ApEn of shoulder joint angle

PAGE 42

42 could possibly indicate the ability of shoulder joint to effectively adapt during movement. Hence, ApEn could be utilized to quanti fy the c omplexity of movements occurring at a particular joint and give a window into the health of the motor system. Complexity can be a periodic or stable, a chaotic, or a completely random state (Stergiou, Harbourne & Cavanaugh, 2006). The periodic or stable s tate is characterized by a completely regular pattern across cycles, exhibiting no variation from this pattern. The periodic or stable state has zero ApEn indicating low complexity (Harbourne & Stergiou, 2009) The random state is characterized by dissimilar pattern s across cycles where the paths are not dependent on each other. This state exhibits relatively greater ApEn suggesting high complexity (Harbourne & Stergiou, 2009) The chaotic state is characterized by a complex, yet organized pattern, with similar paths for each cycle, but not repeating the same path. ApEn values of this signal lies in between the stable and random state suggesting optimal complexity (Harbourne & Stergiou, 2009) Moderate complexity in many biological systems reflects normal healthy function. For instance, greater complexity in the heart and brain waves is correlated with a healthy state (Elbert et al., 1994) An optimal complexity in movement is essential to perform daily function. Harbourne and Stergiou (2009) describe chaos as the representation of healthy and functional movement patterns. Movements after central nervous system damage or dysfunction, for instance in movement disorders such as tardive dyskinesia, show decreased complexity ( v an E mmerik, et al., 1993) Wh ile there has been some investigation into the complexity during walking in the elderly (Kurz & Stergiou, 2003) and infants with developmental delay (Deffeyes, Harbourne, De Jong, et al., 2009; Deffeyes, Harbourne, Kyvelidou,

PAGE 43

43 Stuberg, & Stergiou, 2009) complexity as measured by ApEn has not been explored in UE movements post stroke. Reach to grasp is one of the primary movements performed by the UEs and is often limited aft er stroke ( Woodbury, et al., 2009 ) Therefore, the primary aim of this study was to compare the magnitude of variability and complexity in shoulder, elbow, wrist and proximal interphalangeal (PIP of index finger) flexion/extension joint angles during reach to grasp movements between healthy individuals and individuals with stroke. We hypothesized that the magnitude of variability of shoulder, elbow, wrist and PIP joint of index finger (flexion/extension) angles as measured by SD would be significantly great er post stroke as compared to healthy controls. We also hypothesized that the complexity of shoulder, elbow, wrist and PIP joint of index finger (flexion/extension) joint angles as measured by ApEn would be significantly greater in healthy controls than in individuals with stroke. The secondary aim of this study was to identify the clinical and traditional kinematic correlates of complexity We hypothesized that shoulder, elbow, wrist and PIP joint of index finger ApEn would exhibit significant correlations with Fugl Meyer UE subscale (FM_UE) (Fuglmeyer, Jaasko, Leyman, Olsson, & Steglind, 1975) as well as with conventional measures of UE kinematics. The knowledge gained regarding the nature of the motor impairment m ight be beneficial in designing better UE interventions to enhance adaptability post stroke. Methods Participants The participants were 16 individuals with chronic stroke that were recruited from a larger randomized controlled clinical trial of Constraint Induced Movement Therapy

PAGE 44

44 (CIMT). The protocol of this study was approved by the University of Florida Institutional Review Board and North Florida/South Georgia Veteran Health Systems Research and Development Commit tee. Participants were included if they: (1) were between the ages of 18 90 years of age; (2) had a single ischemic stroke at least 6 months prior; (3) were able to follow two step commands; (4) had no history of more than minor head trauma, subarachnoid h emorrhage, dementia or other neural disorder/dysfunction, drug or alcohol abuse, schizophrenia, serious medical illness, or refractory depression. T able 3 1 provides the demographic details and the FM_UE scores of the participants with stroke. In addition, 9 healthy age matched (within 10 years) controls that were neurologically and orthopedically intact were also r ecruited for the study. Table 3 2 provides the demographic details of the healthy participants Procedures: Set Up and I nstrumentation U pper ext remity (UE) k inematic analysis of reach to grasp Participants grasped a soda can (56 mm in diameter; 208 mm circumference) first with the non paretic UE and then with the paretic UE at the Human Motor Performance Laboratory in the Brain Rehabilitation Res earch Center at the V eteran A ffairs M edical C enter Gainesville, Florida. Grasping a soda can was chosen because it closely approximates the natural and functional use of UE in daily activities. Sixty seven reflective markers were secured to the various la ndmarks of the upper body as seen in figure 3 1. Marker placements were determined using a marker set described by the Plug In UE marker set defined by our laboratory. Participants wore dark colored sleeveless shirts and after placement of the markers, w ere seated on an adjustable, backless bench with knees bent at 90 flexion and feet flat on the floor, hands palm down on the table in front of them and supported in 90 of elbow flexion by arm rests

PAGE 45

45 flush with the table. This position was the starting pos ition for all the trials. A filled soda process to the tip of the middle finger) on a table directly in front of the respective shoulder of the participant. This distance has be (Mark, et al. 1997) Healthy individuals use UE joints alone to reach for objects within this workspace; to obtain objects beyond this boundary; they might involve the trunk by leaning forward. Participants reached for the can, lifted it off the table, and put it bac k down as fast as they could and returned to the starting position. Participants performed command. Participants performed discrete trials because majority of the funct ional tasks K inematics of reaching was recorded using a 12 camera VICON motion capture system (Vicon 612, Oxford Metrics, Oxford, UK) at a sampling frequency of 100 Hz. A preliminary frequency analysis of th e reach to point data across all the conditions indicated that the range of signal frequencies that contain 99.99% of the overall signal power is between 1 and 10 Hz. Therefore, the sampling frequency was set at 100 Hz in order to be at or above a factor o f ten higher than the highest frequency that might contain relevant signal. Midway through the study, the motion capture s oftware was changed from Vicon Workstation t o Vicon Nexus 1.3 (Oxford, UK). This system allowed for higher efficiency, better resoluti on and data within this system were sampled at 200 Hz. All controls and 11 individuals with stroke underwent testing using Vicon Workstation. The remaining five individuals with stroke were tested using Vicon Nexus. Hence, t he data collected using VICON Ne xus was down sampled from 200 to 100Hz for equivalent comparisons

PAGE 46

46 Data processing Data analysis was performed on the last three trials The data were the 3 D positional coordinates of each marker with respect to a laboratory coordinate system throughout t he movement series. The data were then manually labeled, and reconstructed. We analyzed the unfiltered data in order to retain the inherent complexity of the kinematic data (Rapp, Albano, Schmah, & Farwell, 1993) The entire movement cycle was analyzed. The start of reach was identified as the time point at which the velocity of the index finger marker exceeded 5% peak velocity and the termination of reach as the time point at which velocity of this marker fell below 5% peak velocity. One degree of freedom in the sagittal plane (flexion/extension) was used to determine shoulder, elbow wrist and P IP joint angle of index finger. The kinematic data were then modeled using SIMM (4.2, Santa Rosa, CA), which provided the shoulder, elbow, wrist and PIP (index finger) joints of angles. Kinematic variables Four primary kinematic variables wer e computed: peak velocity (PV), time to peak velocity (PVT), trunk displacement (TD) and total movement time (TMT). PV was the highest velocity during the reach, typically occurring at the beginning of the deceleration phase when approaching the target. PV T was calculated from the point of movement onset to peak velocity. TMT was defined as the time from start till end of movement. TD was computed as the total displacement of the trunk in the X, Y and Z positions from start to touch. All the kinematic varia bles were calculated usi ng customized MATLAB scripts

PAGE 47

4 7 Standard d eviation To measure magnitude of variability, SDs of three trials of the shoulder, elbow, wrist and PIP (index finger) joint angles were also computed. Surrogate analysis A surrogation proced ure was applied prior to computing ApEn utilizing Theiler et utilizes a phase randomization t echnique which removes the deterministic structure from the original shoulder, elbow, wrist and PIP (index finger) joint angle time series creating 20 surrogate time series of each trial with the same mean, variance, and power spectrum as the original time series. ApEn was then computed on the original as well as each of the 20 surrogate time series. Significant differences in ApEn between the original and 19 out of 20 surrogate time series would confirm the deterministic nature of the original data. Approx imate e ntropy After verifying that the kinematic data was not random ApEn was obtained using the using MATLAB code (R2009a, Natick, MA) developed by Kaplan and Staffin (1996) utilizing the algorithm provided by (Pincus, Gladstone, & Ehrenkranz, 1991) ApEn is a measure of complexity and determines the randomness in a time series. S houlder, elbow, wrist and PIP (index finger) flexion/extension ang le time series of all three trials were utilized for analysis of ApEn. Each joint angle time series was analyzed from the start of the reach through the entire length of the respective time series including the pauses between the three trials. This procedu re is not the same as computing ApEn on three individual trials of reach to grasp separately. In fact, the pauses between the three

PAGE 48

48 trials were evaluated as part of the temporal structure of the entire time series. Such an approach was adopted because ApEn is effectively a probability measure, which identifies whether small patterns of the time series repeat later in the entire time series. These small patterns might not be repeated in a single trial of reach to grasp movement. This rationale justifies our approach to evaluate the temporal structure of the entire joint angle time series across all trials Hence, four time series were obtained: one for each of the shoulder, elbow, wrist and PIP (index finger) joint angles. The most common method employed in t he computation of ApEn is to identify repeating short patterns of length m across the entire shoulder joint angle time series. Starting with a vector of length m at point pi in the shoulder joint angle time series, the procedure involved counting the numbe r of other vectors at other points pj (j = i) in the same time series which have a similar pattern within r times the standard deviation of the shoulder joint angle time series. As a result, Cm(r), a count of the recurrence of vectors of length m was obta ined. This same procedure was then repeated for all vectors of length m in the shoulder angle time series and summing the logarithm of the results r is a similarity criterion, and provides the limits for assessing the nearness of adjacent data points in the sh oulder joint angle time series. Another parameter, lag, identified the number of time steps between points in one of the length m vectors. Biomechanical data analysis conventionally utilizes r = 0.2 times the standard deviation of the time series, l ag =1 and m = 2 (Slifkin & Newell, 1999) Thereafter, the log of this similarity count Cm(r), was normalized by the number of points in the shoulder angle time series. T he recurrence of vectors of length m + 1 in the entire shoulder joint angle time series was then ob tained

PAGE 49

49 [Cm+1(r)]. ApEn of shoulder angle was thus computed as the natural logarithm of the ratio of Cm(r) and Cm+1(r), as follows: ApEn (X m.r) = log [C m (r)/C m+1 (r)] (3 1) ApEn of elbow, wrist and PIP (index finger) joint angles was also computed i n the same manner. Because the length of the data could affect ApEn values, we normalized the ApEn values of shoulder, elbow, wrist and PIP (index finger) of each participant to their data length and then multiplied the ratio with a constant equal to 100. A more detailed description of the computation of ApEn can be reviewed in the Appendix of Slifkin and Newell (1999). In general, a vector of shorter length repeats more often than a longer one within a time series, thus the lowest possible ApEn value can be the natural logarithm of 1, which is 0, and negative values cannot be obtained. ApEn values range from 0 to 2. In a highly periodic or regular time series, values of Cm(r) can be similar to Cm+1(r) producing ApEn = 0 Hence, smaller values characterize a more regular time series where similar patterns are more likely to follow one another. In contrast, high ApEn values, suggest a highly irregular time series, where the predictability of subsequent patterns is low and ApEn could be close to 2 (Stergiou e t al. 2004). Apart from computing ApEn at various UE joints, we also computed the percentage contribution of each joint to the total ApEn of UE. Total ApEn was computed by adding the shoulder, elbow, wrist and PIP (index finger) t. Thereafter the percentage contribution of shoulder joint was obtained by multiplying the ratio of shoulder ApEn to total ApEn by 100. Similarly, the percentage contribution of elbow joint was computed by multiplying the ratio of elbow ApEn to total ApEn by 100.

PAGE 50

50 The percentage contribution of wrist joint was obtained b y multiplying the ratio of wrist ApEn to total ApEn by 100. And finally, the percentage contribution of PIP joint was calculated by multiplying the ratio of PIP (index finger) ApEn to total ApEn by 100. Such analyses would reveal the distribution of ApEn across the several UE joints. Statistical Analysis Dependent one tailed t tests were conducted to compare ApEn shoulder, elbow, wrist and PIP (index finger) values between the original and surrogate time series using SPSS (17.0, Chicago, IL). For the remaining analyses n on parametric analyses were employed due to the violation of assumptions of normality using SPSS (17.0, Chicago, IL). The first aim of the study was to investigate the differ ences in SD and ApEn shoulder, elbow, wrist and PIP (index finger) between individuals with stroke and healthy controls. For the first aim data were analyzed utilizing the Mann Whitney U test. Data were analyzed at p < 0.05, with Holm s s tep down correctio n procedure. Holm s step down procedure involves sorting the significance values obtained for each hypothesis (here each joint) in ascending order. Consequently, the hypothesis with the lowest significance value was evaluated at p = 0.012 (0.05/4) If the previous hypothesis was significant then the next hypothesis was evaluated at p = 0.016 (0.05/3). Similarly, the next hypothesis was evaluated at p = 0.025 (0.05/2) and the last at p = 0.05. The analysis was not continued further if any hypothesis was reje cted. Further, the percentage contribution of shoulder, elbow, wrist and PIP (index finger) joint ApEn to the total ApEn of UE was also compared between individuals with stroke and healthy controls. This was achieved by utilizing the Mann Whitney U test. The results were analyzed at p < 0.05, with Holm s step down correction procedure as described above

PAGE 51

51 The second aim of the study dealt with identifying clinical and kinematic correlates of ApEn. This was achieved by performing multiple Spearman rank orde r correlation analyses between ApEn shoulder and FM_UE and each of the kinematic variables (MT, PV, PVT, and TD). Similarly, multiple correlation analyses were performed between ApEn elbow and the kinematic variables, and the FM_UE. Multiple correlation an alyses were also performed between ApEn wrist and the kinematic variables, and the FM_UE. Lastly, ApEn PIP was also correlated with FM_UE and each of the kinematic variables. This was a preliminary study to explore the relationship between ApEn of UE joint s and FM_UE and UE kinematic variables post stroke. Hence we did not correct for multiple correlation analyses. Results Dependent one tailed t tests revealed significantly greater shoulder, elbow, wrist and PIP (index finger) joint ApEn in surrogate time s eries for both controls (p<0.05) and stroke (p<0.05) groups. These results suggested that original shoulder, elbow, wrist and PIP (index finger) joint angle time series were deterministic and not derived randomly. The Mann Whitney U test revealed greater S D for shoulder, elbow, wrist and PIP (index finger) angles for individuals with stroke than for healthy controls. However, these differences were not statistically significant (p> 0.012) (Figure 3 2). In contrast, ApEn PIP (index finger) was significantly greater for controls ( Mdn = .66) than for individuals with stroke ( Mdn = .06), U = 0, p < .012, r = .81(Figure 3 3). Similarly, ApEn shoulder was also significantly greater for controls ( Mdn = .14) as compared to individuals with stroke ( Mdn = .03), U = 0 p < .016, r = .81 (Figure 3 3). Likewise, ApEn elbow was significantly greater for controls ( Mdn = .16) than for individuals with stroke ( Mdn = .05), U = 3, p < .025 r = .78 (Figure 3 3). Finally, ApEn wrist was also significantly greater for

PAGE 52

52 controls ( Mdn = .24) than for individuals with stroke ( Mdn = .08), U = 9, p < .05, r = .71 (Figure 3 3). Further, Mann Whitney U test revealed that the percent contribution of ApEn PIP (index finger) joint to total ApEn was significantly greater for controls ( Md n = 52.34) than for individuals with stroke ( Mdn = 29.41), U = 20, p < .012 r = .59 (Figure 3 4). Specifically, ApEn of PIP joint of index finger contributed approximately 52% to total ApEn in healthy controls whereas; PIP joint only contributed 29% to to tal ApEn for individuals with stroke. In contrast, individuals with stroke ( Mdn = 21.32) demonstrated significantly greater percent contribution of ApEn elbow joint to total ApEn than controls ( Mdn = 13.80), U = 19, p < .016, r = .60 (Figure 3 4). In part icular, ApEn of elbow joint contributed approximately 21% to total ApEn for individuals with stroke whereas; elbow joint only contributed 14% to total ApEn for healthy controls. Similarly, percent contribution of ApEn wrist joint to total ApEn was signific antly greater post stroke ( Mdn = 34) as compared to controls ( Mdn = 21.19), U = 29, p < .025, r = .49 (Figure 3 4). Specifically, ApEn of wrist joint contributed approximately 34% to total ApEn for individuals with stroke whereas; wrist joint only contribu ted 21% to total ApEn for healthy controls. However, the difference in percent contribution of ApEn shoulder joint to total ApEn was not significant between controls ( Mdn = 12.67) and individuals with stroke ( Mdn = 15.34), U = 67, ns r = .06 (Figure 3 4). Table 3 3 shows the means and standard deviations of PV, TD, PVT and TMT. ApEn shoulder correlated significantly with FM_UE, PV, TMT and PVT (p< 0.05) but not with TD (p> 0.05) (Table 3 4). Further, significant Spearman correlations were observed between ApEn elbow and TMT and PVT (p< 0.05) and not with FM_UE, PV and TD (p>

PAGE 53

53 0.05) (Table 3 4). ApEn wrist only significantly correlated with TMT (p<0.05) but not with FM_UE, PV, TD and PVT (p>0.05) (Table 3 4). ApEn PIP (index finger) did not significantly corr elate with FM_UE, PV, TD, PVT and TMT (p>0.05) (Table 3 4). Figure 3 5 depict the scatter plots of the correlation between ApEn shoulder and FM_UE, ApEn shoulder and PV, ApEn shoulder and TMT, and ApEn shoulder and PVT. Figures 3 6, 3 7, and 3 8 depict the scatter plots of the correlation between ApEn elbow and PVT ApEn elbow and TM T and ApEn wrist and TMT respectively Discussion The primary purpose of the study was to compare the differences between the magnitude of variability and complexity between ind ividuals with stroke and healthy controls. We employed SD of the shoulder, elbow wrist and PIP (index finger) joint angles to measure the magnitude of variability and ApEn of the same joint angles to quantify the complexity In our study we observed that, although not statistically significant, the magnitude of variability (as measured by SD) was greater in individuals with stroke across all joints as compared to healthy controls. This finding is consistent with the traditional motor control and learning li terature, where variability is considered to be erroneous (Reisman & Scholz, 2006) Individuals living with hemiparesis following stroke demonstrate deficits in motor control, visible as deviations in the kinematics of reach to grasp. Similarly, Woodbury et al. (2009) demonstrated that the SDs of shoulder and elbow joint excursions were greater for individuals with stroke than for healthy c ontrols. Similarly Cirstea and Levin (2000) observed greater SDs of various UE kinematic parameters such as movement time, end point and trunk tangential velocity utilizing a reach to point task paradigm post stroke. Thus, our findings are consistent with the literature.

PAGE 54

54 In contrast, complexity across all joints as measured by ApEn was significantly greater for healthy controls as compared to individuals with stroke. Interestingly, further analysis revealed that complexity was significantly greater in the PIP joint of the index finger than in the shoulder, elbow and wrist joints in healthy controls. This unique finding might imply that healthy controls made fewer adjustments with the proximal joints (especially shoulder) suggesting that shoulder was primari ly utilized for stabilization of the arm during the reach to grasp task. On the other hand, the PIP joint of the index finger might have produced greater adjustments essential in manipulating the grasp around the can during the reach to grasp task. These f indings are consistent with the current literature, which support the versatile nature of hand (Lemon, 1993; Tallis, 2003). The advanced ability of the hand to grasp and manipulate objects of various sizes, shapes and textures is one of the key features of the human motor system (Begliomini, Nelini, Caria, Grodd, & Castiello, 2008) In fact, there is ample evidence supporting that the kinematics of end effector (i.e. hand) is the primary variable controlled during movement (Todorov, Shadmehr, & Bizzi, 1997) For instance, kinematics characteristics (trajectory of han d path and bell shaped velocity profile) of simple reaching movements are retained despite changes in visual feedback (Thach, Goodkin, & Keating, 1992; Wolpert, Ghahramani, & Jordan, 1995) and dynamics of the envir onment (Lacquaniti, Soechting, & Terzuolo, 1986; Shadmehr & Mussa Ivaldi, 1994) These studies indicate that central nervous system utilizes the kinematics of end effector to plan and execute the movement (Winstein Wing, & Whitall, 2003). Greater complexity across all UE joints imparts adaptability in the motor system to adapt to common stresses encountered in daily life (Harbourne & Stergiou, 2009) From

PAGE 55

55 a motor control perspective, optimal complexity allows an individual to explore the redundant degrees of freedom of the motor system. For instance, a healthy individual would be able to effectively reach and grasp objects of different sizes placed at varying locations. I n contrast, less complexity suggests that the system is more rigid, (less flexible) as evident in individuals with stroke (Harbourne & Stergiou, 2009; Scholz, 1990) Several changes associated with stroke, such as, spasticity, decreased range of motion ( Cirstea & Levin, 2000) difficulty dealing with interaction torques produced by muscle contractions, and abnormal motor recruitment patterns (Dewald, Pope, Given, Buchanan, & Rymer, 1995) could possibly reduce the a bility of the motor system to effectively adapt to the environment. For instance, individuals living with hemiparesis following stroke often demonstrate stereotypical movement patterns limiting their repertoire of behaviors (Cirstea & Levin, 2000; Michaelsen, Jacobs, Roby Brami, & Levin, 2004) This could possibly explain a decline in shoulder, elbow and wrist joints in individuals with stroke as compared to healthy adults. Further, individuals with stroke also exhi bited significant reduction in ApEn of PIP joint of index finger. These findings are consistent with other studies demonstrating slow and less accurate finger and hand movements; poor modulation of fingertip forces and decreased ability to move fingers ind ividually (Cruz, Waldinger, & Kamper, 2005; Grichting, Hediger, Kaluzny, & Wiesendanger, 2000; Hermsdorfer, Hagl, Nowak, & Marquardt, 2003; Lang & Schieber, 2003) Significant reduction in ApEn of PIP joint of inde x finger could possibly be explained by the fact that the motor neuron pools of distal UE segments are primarily innervated by the corticospinal tract which is frequently compromised in stroke (Colebatch & Gandevia, 1989)

PAGE 56

56 Additionally, the percent contributio n of ApEn PIP (index finger) joint to total ApEn was significantly greater for controls than individuals with stroke. In contrast, individuals with stroke demonstrated significantly greater percent contribution of ApEn wrist and elbow joints to total ApEn than controls. This finding suggests that individuals with stroke possibly made significantly greater adjustments with wrist and elbow and significantly fewer adjustments with the PIP joint of the index finger than controls during the reach to grasp task. Such findings imply that individuals with stroke potentially utilized an alternative compensatory strategy to accomplish the reach to grasp task. Compensatory strategies during grasping post stroke have been reported in the literature. For instance, Raghav an, Santello, Gordon and Krakauer (2010) observed increased flexion at the metacarpophalangeal joint than PIP joint of the fingers while grasping concave and convex shaped objects post stroke. Greater complexity at joints other than PIP might be an altern ative strategy to explore movement redundancy in the more affected UE to accomplish the task goal. Understanding the phenomenon of movement redundancy i.e. how multiple effectors coordinate to produce a goal directed movement still remains a challenge to m otor control researchers (Diedrichsen, Shadmehr, & Ivry, 2009) In particular, the problem of motor coordination deals with how work is distributed across multi ple effectors (muscles, joints) when multiple options exist to perform a task, commonly referred to as the degrees of freedom problem (Bernstein, 1967). Optimal control theory suggests that an optimization process for the tasks might be a potential solutio n to the degree of freedom problem of motor control (Diedrichsen, et al., 2009) Optimal control theory proposes that selection of effectors for a particular ta sk is the consequence of an

PAGE 57

57 optimization process based upon the cost function of the goal and effort required to accomplish the goal. Based upon this theory, one could then propose that stroke might change the cost function. For individuals who have modera te UE deficits post stroke, such as our participants who had some ability to voluntarily extend the fingers, manipulating the PIP joint of index finger around the can could potentially require too much effort; the compensation strategy involving the wrist and elbow joints might result in reoptimization on the basis of a new cost function. In other words, individuals with stroke might have utilized the wrist and elbow joints while exploiting the redundancy in order to accomplish the task goal through redistr ibution of work across effectors. In fact, using the wrist, in particular, may have made it easier to open and close the fingers due to the biomechanical properties of the long flexors (flexor digitorum superficialis) which cross both the wrist, and finge rs. Decrement in complexity is not only associated with pathology like stroke, but also in older adults For example: Sosnoff and Newell, ( 2006b ) observed that older adults demonstrate reduced ApEn of finger isometric force production. Thus, a decline in complexity could be characteristic of aging and pathology such as stroke, which limits the ability of the motor system to effectively adapt to the changes in the envi ronment. A second purpose of the study was to investigate the clinical correlates of complexity of shoulder, elbow, wrist and PIP (index finger) joint angles. This objective was achieved by correlating ApEn shoulder, elbow, wrist and PIP (index finger) wi th FM_UE and other UE conventional kinematic measures. ApEn shoulder demonstrated a significant positive correlation with FM_UE and PV. These findings indicate that participants with stroke with greater FM_UE scores (or less motor impairment) or

PAGE 58

58 greater PV might exhibit greater complexity at the shoulder joint. Further, a significant negative correlation was observed between ApEn shoulder and PVT and ApEn elbow and PVT suggesting that participants with stroke with greater complexity at shoulder and elbow jo int s attained peak velocity faster as compared to individuals with less shoulder and elbow ApEn. Significant negative correlations were also observed between ApEn shoulder, elbow and wrist with TMT, indicating that participants with stroke with greater com plexity at shoulder, elbow and wrist joints required less time to accomplish the reach grasp movement. One must not make causal inferences from correlational data. Nonetheless, our data suggest that greater complexity is associated with better movement qua lity in these individuals. On the other hand, both approximate entropy and the other kinematic variables may simply reflect an impaired motor system. The presence of significant correlations might suggest that measuring both the kinematic variables and app roximate entropy is redundant. However, until there is greater understanding of complexity in UE movements, it would be premature to suggest such redundancy. Further research is necessary to understand specific neurological mechanisms contributing to the d ecline in complexity in UE joints post stroke compared to other kinematic and functional variables and to determine the effects of intervention on these variables. In particular, the effects of location and size of brain lesion, severity of the lesion, int egrity of the descending motor pathways, individual degree of spontaneous recovery, and the duration of stroke onset upon complexity in UE joints post stroke need to be explored. There are certain limitations of this study. Given the heterogeneity observ ed in stroke, this sample size was relatively small. The lack of significant differences between

PAGE 59

59 groups in shoulder and elbow SD might reflect a lack of statistical power. Additionally, no significant correlations between ApEn PIP of index finger and FM_UE and measures of UE kinematics might also indicate a lack of statistical power. Implications for Rehabilitation This study was the first attempt to delineate the differences between the two dimensions of variability in shoulder, elbow, wrist and PIP (inde x finger) joint angles post stroke. Specifically, the magnitude of variability and complexity exhibited an inverse relationship. As hypothesized, although not significant, the magnitude of variability in the shoulder, elbow, wrist and PIP (index finger) jo int angles was increased post stroke. However, complexity was decreased in the shoulder, elbow, wrist and PIP (index finger) joint angles post stroke. ApEn seems to be a more sensitive measure since, e ven with a small cohort of the stroke population we wer e able to significantly differentiate between healthy controls and individuals with stroke utilizing non linear measure of variability, such as ApEn. Contrarily, employing linear measure of variability, such as SD failed to detect differences between heal thy controls and individuals with stroke. Additionally, a decrement in complexity post stroke indicated the compromised adaptability of the motor system. It might then seem intuitive to refer the complexity of the system as Most impo rtantly, this study provides preliminary evidence that variability is not always negative. An optimal amount of adaptive variability provides individuals with a repertoire of movement strategies to accomplish a task. Thus, rehabilitation therapists should aim to augment adaptive variability post stroke. However, it would be difficult to measure ApEn in the clinic at the present time. T here have been no investigations into which interventions facilitate increased UE adaptive variability and which do not.

PAGE 60

60 Nev ertheless, few suggestions could be incorporated into existing treatment protocols post stroke. Interventions could be designed in order to promote the adaptive capacity of individuals with stroke. The goal of intervention might be to enh ance the complexit y of movement. Incorporating multiple experiences in the therapeutic milieu might provide a rich and complex repertoire of behavioral strategies facilitating the adaptive variability post stroke. Specifically, participants could practice functional tasks a midst several task constraints including biomechanical, cognitive, and personal during interventions. For example, grasping and lifting a can could be practiced from surfaces of varying heights. Similarly, the amount of force required for grasping might be altered as well, where individuals practice grasping objects of different weights, sizes, and shapes. Constantly challenging the participation of clients during therapy sessions could maintain them in a state of chaos (described earlier) reflecting optima l variability These strategies might allow individuals with stroke to explore several problem solving strategies during task performance resulting in an optimal strategy to perform functional task s. Future research is necessary to determine the reliabilit y of ApEn in UE kinematics. In conclusion variability is inherent in UE movements. Linear measures of variability consider increase in movement variability as noise However, non linear measures reveal the hidden complexity Individuals with stroke might exhibit greater magnitude of variability and lower complexity joints. Further research is warranted to generalize the use of non linear measures to other neurologically impaired populations, apart from stroke.

PAGE 61

61 T able 3 1 Participant demographics: individuals with stroke Participant Gender Age Affected Side UE_FM Lesion Location Months after CVA 1 M 76 L 41 Right middle cerebral artery 102 2 M 62 L 46 Right M1, middle cerebral artery 48 3 F 70 L 44 Right Stri atoscapular infarct 131 4 F 66 R 58 Left Midlle/Posterior cerebellar artery 102 5 M 73 R 45 Left medullary/brainstem infarct 103 6 M 76 R 53 Left middle cerebral artery 174 7 M 55 L 41 Right medial medullary infarct 43 8 M 66 L 43 Right Striatoscapula r infarct 105 9 F 47 L 35 Right basal ganglia 7 10 M 62 L 27 Right middle cerebral artery 118 11 F 64 L 31 Right lacunar infarct 67 12 F 62 L 27 Posterior ventricular white matter 19 13 M 72 R 38 Left lacunar infarct 24 14 M 77 R 38 Left pontine infa rct 34 15 M 72 L 30 Right middle cerebral artery 48 16 M 68 R 31 Left middle cerebral artery 16 Mean ( SD) 66.75 ( 8.1) 39.25 (8.9) 71.3 ( 48.8)

PAGE 62

62 Table 3 2 Participant demographics: healthy controls Participant Gender Age Dominant Hand 1 F 61 R 2 F 43 R 3 F 51 R 4 M 62 L 5 F 62 R 6 F 56 R 7 F 58 R 8 F 65 R 9 F 57 R Mean ( SD) 57.2 ( 6.7) Table 3 3. Mean of the conventional upper extremity (UE) kinematics measures of participants Mean (SD) PV (meters/second) TD (millimeter) PVT (s econds) TMT (seconds) Stroke 0.57 (0.18) 141 (51) 0.50(0.10) 1.44(0.60) Control 0.95 (0.14) 44.62 (16.90) 0.21 (0.01) 0.51 (0.10) PV: Peak velocity; TD: Trunk displacement; PVT: Time to peak velocity; TMT: Total movement time. Table 3 4. Spea rma n Rank correlations between approximate entropy (ApEn) of upper extremity joints and kinematic variables of reach to grasp. ApEn PV TMT PVT TD FM_UE Shoulder .60 .81 .70 .40 .60* Elbow .11 .60 .50* .32 .20 Wrist .09 .50* .35 .15 .03 PIP .15 .30 .25 .40 .37 ApEn Approximate entropy; PV Peak velocity; TMT Total movement time; PVT Time to peak velocity; TD Trunk displacement; FM_UE Fugl Meyer UE subscale; PIP Proximal interphalangeal joint. *= Significant at p < 0.0 5.

PAGE 63

63 Figure 3 1. U pper extremity (UE) marker set. Figure 3 2. S tandard deviation (S D ) of various UE joints between healthy controls and individuals with stroke

PAGE 64

64 Figure 3 3. Approximate Entropy ( ApEn ) of various UE joints between healthy controls and individuals with st roke (* = significant )

PAGE 65

65 Figure 3 4 Approximate entropy (ApEn) percent of each joint to total ApEn in healthy controls and individuals with stroke (* = significant)

PAGE 66

66 CHAPTER 4 DOES MOVEMENT SPEED AND/OR RHYTHM IMPROV E U PPER E XTREMITY ADAPTIVE VARIABILIT Y POST STROKE? B ackground Stroke is one of the leading causes of adult onset disability (Duncan, 1995). Every year over 795,000 individuals are affected by stroke in United States (American Heart Association, 2010). Up to 85% of individuals with stroke e xhibit upper extremity (UE) paresis immediately post stroke (Olsen, 1990). Despite rehabilitation, residual UE impairments still exist (Duncan, et al., 2000; Nakayama, et al., 1994) Hence, there is a need for more effective treatments. Discovering practice variables that promote better motor control may assist in the discovery of more effective treatment. One of the most common motor impairments exhibited after stroke is stereotypical movement patterns (Cirstea & Levin, 2000) According to Dynam ical systems theory because individuals with stroke easily fall into the pattern and return to that pattern even when perturbed or interrupted (Kamm, et al., 1990) We discovered that these stereotypical patterns are characterized with low complexity in shoulder an d elbow flexion as measured by a pproximate e ntropy (ApEn) (Sethi, Patterson, McGuirk, & Richard s, 2009). This is problematic for functional use of the UE because, as proposed by DST, greater complexity provides individuals with a repertoire of movement strategies to adapt movement patterns and successfully meet the demands of everyday changing task s (Scholz, 1990) For example, reaching for a soda can on a low surface would employ different movement strategies than reaching overhead for the same can. Additionally, the repeated performance of movements in an abnormal manner may

PAGE 67

67 result in pa in and contractures of various UE joints (Cirstea & Levin, 2000) which might further limit the adaptive variability of UE post stroke. Because of reduced adaptive variability after stroke, the DST perspective would suggest that a goal for therapy may be to increase adaptive variability i n individuals with stroke to enhance UE motor control and function. Utilizing the principles of DST, adaptive variability might be enhanced by changing certain task constraints, known as control parameters (Newell, 1986). Task constraints or control param eters are variables that might be highly specific, such as myelination, particular muscle strength, movement speed or nonspecific, such as emotional or motivational aspects. Changing the appropriate task constraints might transition the stable or stereotyp ical movement patterns with low adaptive variability to more adaptable ones. The next logical step then is to identify task constraints that might be potential control parameters. Movement speed has been shown to drive the system from one stable state (or movement pattern) to another. For instance, Kelso (1984) utilized a bimanual coordination paradigm to demonstrate the effect of speed upon the coordination between the right and left index fingers in healthy participants. Participants were asked to maintai n a rhythmic anti phase pattern, characterized by abduction of one index finger, while adduction of the other, synchronized with the beat of a metronome. As the frequency of metronome increased from 1.25 to 3.5 hertz, participants transitioned from the ant i phase pattern into an in phase pattern characterized by simultaneous abduction and adduction of both index fingers. Similar observations have been noted for gait patterns as well (Diedrich & Warren, 1995) Normal gait speed is between 1.2 1.5 meter s/second, when the speed is increased to

PAGE 68

68 2.1 meters/second, gait transitions from walking (stable state) to running (another stable state) (Diedrich & Warren, 1995) The effect of movement speed upon UE adaptive variability has not been studied. How ever, the effect of movement speed upon the adaptive variability of certain kinematic parameters of gait has been studied. For instance, Buzzi and Ulrich, ( 2004 ) demonstrated the effect of speed on A pEn of thigh, shank and foot segments in 8 and 110% of the self selected speed on a treadmill. They discovered that ApEn of thigh, shank and foot decreased as a function of in creasing speed in both children with and typical children. Because ApEn should be lower in rhythmic tasks, such as walking, in a healthy nervous system (Vailliancourt and Newell, 2002), this reduction in adaptive variability suggests that f aster speed drove the motor system to a more normal state in these children. However, Vailliancourt and Newell (2002) proposed that increased ApEn reflects a healthy state of the motor system in fixed point tasks, such as goal directed reaching in healthy individuals. It w ould then seem intuitive, that greater speed might drive the UE motor system of individuals post stroke to output movements with greater ApEn during reaching movements. Apart from movement speed, rhythmic cues could also act as a potentia l task constraint to augment adaptive variability of UE post stroke. The structured time information in auditory rhythm cues enhances the spatial temporal characteristics of movement by entraining the timing of muscle activation patterns (Thaut, Kenyon, Schauer, & McIntosh, 1999; Thaut, et al., 1996) Movements made with rhythmic cues also add stability to motor control. Several studies (McIntosh, Brown, Rice, & Thaut,

PAGE 69

69 1997; Thaut, et al., 2002; Thaut, McIntosh, & Rice, 1997; Thaut, et al., 1996) demonstrated that movements produced to auditory rhythmic cuing are more accurate, faster and have a smaller magnitude of variability, than those produced without such cues. Thaut, M cIntosh, Rice and Prassas (1993) demonstrated the immediate entrainment effects of auditory rhythmic cuing on gait post stroke. Ten individuals with hemiparesis walked with rhythmic cues, where the cue frequency was matched to the step rate of the particip ants. Thaut et al. (1993) observed a significant improvement in weight bearing stance time on the paretic leg, and stride symmetry. The motor unit recruitment patterns also improved as evident from the reduction of magnitude of variability in the electomyo graphic patterns of gastrocnemius muscle bilaterally. Rhythmic entrainment also occurs for UE movements. Thaut, et al., (2002) observed immediate entrainment effects on reaching post stroke. Twenty one individuals with hemiparesis performed sequential rea ching trials for 30 seconds with and without auditory rhythmic cuing. Reaches made with auditory cues exhibited better movement composition; reduced trunk displacement and increased shoulder and elbow joint excursions, straighter reaching trajectories. In addition, the magn itude of variability of timing was also reduced in reaches made with auditory rhythmic cues. These findings suggest that reaches made with auditory rhythmic cues may decrease the magnitude of variability of movements post stroke. Whether such cues change adaptive variability in reaching po st stroke has not been tested. The magnitude of variability quantifies the amount of error associated with kinematics of the movement under study. Adaptive variability provides information about the comp lexity of movement Specifically, measures of adaptive variability unfold the underlying time

PAGE 70

70 dependent fluctuations in a variable over time. Measures of c omplexity utilize the entire range of values of a movement parameter as opposed to magnitude. We know that magnitude of variability and adaptive variability most often exhibit an inverse relationship in UE movements (Harbourne & Stergiou, 2009) We showed that shoulder elbow wrist and PIP (index finger) angles in reach to grasp movements exhibited greater magnitude of variability and decreased adaptive variability post stroke as compared to healthy participants (Sethi, et al., 2009). Because of the inverse relationship between magnitude of variability and adaptive variability it would then seem intuitive that reaches made with auditory rhythmic cues might enhance the adaptive variability of UE. Auditory rhythmic cuing could possibly act as an appropriate task constraint and transition the UE movement from stereotyp ical to more adaptable movement patterns post stroke. The purpose of this study was to investigate whether the adaptive variability of UE, as measured by ApEn, is enhanced when reaching naturally at a fast speed and/or to auditory rhythmic cues compared t o non cued movements made at a self selected speed in individuals with stroke. We hypothesized that individuals with stroke would exhibit significantly greater shoulder, elbow, wrist and proximal interphalangeal (PIP of index finger) flexion/extension join t angles ApEn while reaching at a faster speed and/or to auditory rhythmic cues versus reaching at comfortable pace.

PAGE 71

71 Methods Participants Thirteen individuals with chronic stroke (mean age: 67 8.9) were recruited from a larger randomized controlled clini cal trial. The protocol of this study was approved by the University of Florida Institutional Review Board and North Florida/South Georgia Veteran Health Systems Research and Development Committee. Participants were included if they: (1) were between the a ges of 18 90 years of age; (2) had a single ischemic stroke at least 6 months prior; (3) were able to follow two step commands; (4) had no history of more than minor head trauma, subarachnoid hemorrhage, dementia or other neural disorder/dysfunction, drug or alcohol abuse, schizophrenia, serious medical illness, or refractory depression. Table 4 1 provides the demographic details and the Fugl Meyer UE subscale scores of the participants with stroke. In addition, 8 healthy age matched (within 10 years) contr ols that were neurologically and orthopedically intact were also r ecruited for the study. Table 4 2 provides the demographic details of the healthy participants. Procedures: Set U p and I nstrumentation U pper extremity (UE) k inematic analysis of reach to gra sp Participants performed reach to point movements with both UE at the Human Motor Performance Laboratory in the Brain Rehabilitation Research Center Veteran Affairs Medical Center Gainesville, Florida. Sixty seven reflective markers were secured to the various landmarks of the upper body as seen in Figure 4 1. Marker placements were determined using the Plug In UE marker set defined by our laboratory. P articipants were seated on an adjustable, backless bench with knees bent at 90 flexion and feet flat on the floor, hands palm down on the table in front of them

PAGE 72

72 and supported in 90 of elbow flexion by arm rests level with the table. Participants acromion process to the tip of the middle finger), at midline on a table directly in front of them. Healthy individuals use UE joints alone to reach for objects within this workspace; to obtain objects beyond this boundary, they typically involve anterior flexion of the trunk (Ma initially at their comfortable pace, followed by as fast as possible. Lastly participants performed four continuous trials paced with a metronome matched to the comfortabl e pace produced by the paretic UE. The first reach of each trial was discarded as practice and data analysis was completed on the remaining 3 reaches. Kinematics of reaching were recorded using a 12 camera VICON motion capture system ( Vicon 612, Oxford Met rics, Oxford, UK) at a sampling frequency of 100 Hz. A preliminary frequency analysis of the reach to point data across all the conditions indicated that the range of signal frequencies that contain 99.99% of the overall signal power is between 1 and 10 Hz Therefore, the sampling frequency was set at 100 Hz in order to be above a factor of ten higher than the highest frequency that might contain relevant signal. M idway through the study, the motion capture software was changed from Vicon Workstation t o Vic on Nexus 1.3 (Oxford, UK). This system allowed for better resolution and data within this system were sampled at 200 Hz. All controls and 8 individuals with stroke underwent testing using Vicon Workstation. The remaining five individuals with stroke were t ested using Vicon Nexus. Hence, t he data collected using VICON Nexus were down sampled from 200 to 100Hz to produce equivalent comparisons

PAGE 73

73 Data processing Data analysis was performed on the last three trials. The data were the 3 D positional coordinates of each marker with respect to a laboratory coordinate system throughout the movement series. The data were then manually labeled, and reconstructed. We analyzed the unfiltered data in order to retain the inherent structure of variability of the kinematic data (Rapp et al., 1993). The entire movement cycle was analyzed. The start of reach was identified as the time point at which the velocity of the index finger marker exceeded 5% peak velocity and the termination of reach as the time point at which veloci ty of this marker fell below 5% peak velocity. One degree of freedom in the sagittal plane (flexion/extension) was used to determine shoulder, elbow, wrist and P IP joint angle of index finger. The kinematic data were then modeled using SIMM (4.2, Santa Ros a, CA), which provided the shoulder, elbow, wrist and PIP (index finger) joints of angles. Surrogate analysis A surrogation procedure was applied prior to computing ApEn utilizing Theiler et e deterministic (non phase randomization technique which removes the deterministic structure from the time series creating 20 surrogate time series of each trial with the same mean, variance, and power spectrum as the original time series. ApEn was then computed on the original as well as each of the 20 surrogate time series. Significant differences in ApEn between the original and 19 out of 20 surrogate time series wo uld confirm the deterministic nature of the original data.

PAGE 74

74 Approximate e ntropy After verifying that the kinematic data was not random ApEn was obtained using the using MATLAB code (R2009a, Natick, MA) developed by Kaplan and Staffin (1996) utilizing the algorithm provided by (Pincus, et al., 1991) ApEn is a measure of complexity and determines the randomness in shoulder and elbow joint angle t ime series. S houlder, elbow, wrist and PIP (index finger) flexion/extension angle time series of all three trials were utilized for analysis of ApEn. Each joint angle time series was analyzed from the start of the reach through the entire length of the res pective time series including the pauses between the three trials. This procedure is not the same as computing ApEn on three individual trials of reach to grasp separately. In fact, the pauses between the three trials were evaluated as part of the temporal structure of the entire time series. Such an approach was adopted because ApEn is effectively a probability measure, which identifies whether small patterns of the time series repeat later in the entire time series. These small patterns might not be repea ted in a single trial of reach to grasp movement. This rationale justifies our approach to evaluate the temporal structure of the entire joint angle time series across all trials Hence, four time series wer e obtained: one for each shoulder elbow wrist a nd PIP (index finger) joint angles. The most common method employed in the computation of ApEn is to identify repeating short patterns of length m across the entire shoulder joint angle time series. Starting with a vector of length m at point pi in the sho ulder joint angle time series, the procedure involved counting the number of other vectors at other points pj (j = i) in the same time series which have a similar pattern within r times the standard deviation of the shoulder joint angle time series. As a result, Cm(r), a count of the recurrence of vectors of length m was obtained. This same procedure was then repeated for all

PAGE 75

75 vectors of length m in the shoulder angle time series and summing the logarithm of the results r is a similarity criterion, and pr ovides the limits for assessing the nearness of adjacent data points in the sh oulder joint angle time series. Another parameter, lag, identified the number of time steps between points in one of the length m vectors. Biomechanical data analysis convention ally utilizes r = 0.2 times the standard deviation of the time series, lag =1 and m = 2 (Slifkin & Newell, 1999) Thereafter, the log of this similarity count, Cm(r), was normalized by the number of points in the shoulder angle time series. Following this, the recurre nce of vectors of length m + 1 was obtained [Cm+1(r)], in the entire shoulder joint angle time series ApEn of shoulder angle was thus computed as the natural logarithm of the ratio of Cm(r) and Cm+1(r), as follows: ApEn (X m.r) = log [C m (r)/C m+1 (r)] (4 1) ApEn of elbow, wrist and PIP (index finger) joint angles were also computed in the same manner. Because the length of the data could affect ApEn values, we normalized the ApEn values of shoulder, elbow, wrist and PIP (index finger) of each partic ipant to their data length and then multiplied the ratio with a constant equal to 100. A more detailed description of the computation of ApEn can be reviewed in the Appendix of Slifkin and Newell (1999). In general, a vector of shorter length repeats more often than a longer one within a time series, thus the lowest possible ApEn value can be the natural logarithm of 1, which is 0, and negative values cannot be obtained. ApEn values range from 0 to 2. In a highly periodic or regular time series, values of C m(r) can be similar to Cm+1(r) and ApEn = 0. Hence, smaller values characterize a more regular time series where similar patterns are more likely to follow one another. In contrast, high

PAGE 76

76 ApEn values, suggest a highly irregular time series, where the predic tability of subsequent patterns is low and ApEn could be close to 2 (Stergiou et al. 2004). Apart from computing ApEn at various UE joints, we also computed the percentage contribution of each joint to the total ApEn of UE. Total ApEn was computed by add ing the shoulder, elbow, wrist and PIP (index finger) Thereafter the percentage contribution of shoulder joint was obtained by multiplying the ratio of shoulder ApEn to total ApEn by 100. Similarly, the percentage contribution o f elbow joint was computed by multiplying the ratio of elbow ApEn to total ApEn by 100. The percentage contribution of wrist joint was obtained b y multiplying the ratio of wrist ApEn to total ApEn by 100. Finally, the percentage contribution of PIP joint w as calculated by multiplying the ratio of PIP (index finger) ApEn to total ApEn by 100. Such analyses would reveal the distribution of ApEn across the several UE joints. Statistical Analysis Dependent one tailed t tests were conducted to compare ApEn sho ulder, elbow, wrist and PIP (index finger) values between the original and surrogate time series using SPSS (17.0, Chicago, IL) Non parametric analyses were employed due to the violation of assumptions of normality using SPSS (17.0, Chicago, IL). Wilcoxon Signed Rank tests were employed to differentiate ApEn shoulder, elbow, wrist and PIP (index finger) between the comfortable pace and fast and comfortable pace and metronome reaching conditions within controls and individuals with stroke. Data were analyze d at p < 0.05, with Holm s step down correction procedure. Holm s step down procedure involves sorting the significance values obtained for each hypothesis in an ascending order. Consequently, the hypothesis with the lowest significance value was evaluated at p = 0.006 (0.05/8) If the previous hypothesis was significant then the next hypothesis was

PAGE 77

77 evaluated at p = 0.007 (0.05/7). Similarly, the next hypothesis was evaluated at p = 0.008 (0.05/6), followed by p = 0.01 (0.05/5), then at p = 0.012 (0.05/4), then at p = 0.016 (0.05/3), followed by p = 0.025 (0.05/2) and the last at p = 0.05. The analysis was not continued further if any hypothesis was rejected. Further, the percentage contribution of shoulder, elbow, wrist and PIP (index finger) joint ApEn to the total ApEn of UE was also compared between comfortable pace reaching conditions of controls and individuals with stroke utilizing the Mann Whitney U test The results were analyzed at p < 0.05, with Holm s step down correction procedure as described a bove Similar comparisons were also made in fast reaching as well as metronome reaching conditions separately. If the percentage contribution of UE joints ApEn to total ApEn was not significantly different between comfortable pace reaching conditions of co ntrols and individuals with stroke further analysis involving fast and metronome reaching conditions was not performed. Results Dependent one tailed t tests revealed significantly greater shoulder, elbow, wrist and PIP (index finger) joints ApEn in surrog ate time series than original time series for both controls (p<0.05) and stroke (p<0.05) groups. These results suggested that original shoulder, elbow, wrist and PIP (index finger) joint angle time series were deterministic and not derived randomly. For he althy controls, Wilcoxon Signed rank test revealed greater ApEn in shoulder, elbow, wrist and PIP (index finger) joints in the fast reaching and metronome conditions as compared to the comfortable pace reaching condition (figure 4 2). However, t he lowest p value obtained was not statistically significant at the corrected level of significance (p > 0.006). Based upon Holm s step down procedure further analyses were not conducted. Thus, ApEn in shoulder, elbow, wrist and PIP

PAGE 78

78 (index finger) joints was not sign ificant ly different between comfortable pace and fast and comfortable pace and metronome reaching conditions in healthy controls. For participants with stroke, Wilcoxon Signed rank test revealed significantly greater ApEn shoulder in the metronome conditio n ( Mdn =0.18) as compared to the comfortable pace condition ( Mdn =0.05), T =0, p < .006, r = 0.70 (figure 4 3). ApEn shoulder was also significantly greater when reaching faster ( Mdn =0.10) in comparison to the comfortable pace condition ( Mdn =0.05), T =0, p < .007, r = 0.70 (figure 4 3). Similarly, Wilcoxon Signed rank test further demonstrated significantly greater ApEn elbow in the metronome condition ( Mdn =0.20) as compared to the comfortable pace condition ( Mdn =0.05), T =0, p < .008, r = 0.70 (Figure 4 3) and the reaching faster condition ( Mdn =0.12) in comparison to the comfortable pace condition ( Mdn =0.05), T =0, p < .01, r = 0.70 (Figure 4 3). In addition, Wilcoxon Signed rank test also demonstrated significantly greater ApEn wrist in the metronome condition ( Mdn =0.47) as compared to the comfortable pace condition ( Mdn =0.10), T =0, p < .012, r = 0.70 (Figure 4 3) and when reaching faster ( Mdn =0.30) in comparison to the comfortable pace condition ( Mdn =0.10), T =0, p < .016, r = 0.70 (Figure 4.5). Lastly, ApEn PIP joint of index finger was significantly greater in the metronome condition ( Mdn =0.58) as compared to the comfortable pace condition ( Mdn =0.16), T =7 p < .025, r = 0.54 (Figure 4 3 ), but not significantly greater (p > 0.05) when reachin g faster in comparison to the comfortable pace condition (Figure 4 3). Further, the percentage contribution of ApEn between controls and individuals with stroke in comfortable pace reaching condition revealed that none of the UE joints attained t he lowes t p value at the corrected le vel of significance (p > 0.012) (Figure s 4 4

PAGE 79

79 and 4 5 ). Therefore, based upon Ho l m s procedure further percentage contribution comparisons in fast pace and metronome reaching conditions between controls and individuals with stro ke were not conducted. Discussion The main objective of this study was to investigate the immediate effect of reaching faster or with auditory rhythmic cues on the adaptive variability of UE as measured by ApEn in healthy controls as well as individuals wi th stroke. The adaptive variability in shoulder, elbow, wrist and PIP (index finger) joints was not significantly different across the three reaching conditions in healthy controls. Healthy controls might already exhibit an optimal level of adaptive variab ility; hence changing the reaching condition did not a ffect UE adaptive variability. The effect of movement speed on UE adaptive variability had not previously been studied. However, speed has been shown to influence adaptive variability in gait: increasi ng speed decreases the adaptive variability of certain kinematic gait parameters (Buzzi & Ulrich, 2004) In contrast, our findings run counter to those observed in gait studies (Buzzi & Ulrich, 2004) : faster reaching movements revealed greater adaptive variability. This is consistent with postulates proposed by Vailliancourt and Newell (2002) who suggested that greater ApEn during reaching and l ower ApEn while walking characterize healthy motor behavior. Thus, our findings indicate that making individuals with stroke reach faster moves the UE motor system towards a more normal state as reflected in greater ApEn. However, the underlying neurologic al mechanisms responsible for enhancing UE adaptive variability as a function of increase in speed are still not clear.

PAGE 80

80 Apart from movement speed we also discovered immediate positive effects of auditory rhythmic stimulation in enhancing the adaptive vari ability of UE post stroke. Our findings are consistent with the existing literature supporting the positive influence of auditory rhythmic cueing on reaching post stroke. Thaut, et al., (2002) observed that auditory rhythm cues immediately stabilized the k inematics of reaching with low variability of movement time as well as reaching trajectories. One of the possible reasons for the augmentation of ApEn with auditory rhythm cueing could be attributed to the interaction of the auditory and motor systems (Thaut, et al., 1999) Auditory signals are known to raise the excitability of spinal motor neurons mediated by the auditory motor circuitry at the reticulo spinal level (Rossignol & Jones, 1976) Rhythmic auditory stimulation, allows the brain to map and scale smoother temporal parameters of changes in position of the paretic arm throughout the entire movemen t cycle (Thaut, et al., 2002). Thaut, Kenyon, Schauer and McIntosh, (1996) suggested that external auditory cues provide a temporal constraint to the reaching movement. Once the temporal constraint is added to the movement; the rhythmic cue acts as an exter nal forcing function and simplifies the motor tas k to reach between two targets. Moreover the structured time information in auditory rhythm cues enhances the spatial temporal characteristics of movement by entraining the timing of muscle activation patte rns as measured by electromyography (Thaut, et al., 1999; Thaut, et al., 1996) The entrainment of the timing of muscle activation patterns due to the auditory rhythm cues might be one of the mechanisms to facilita te the adaptive variability of UE post stroke. Thus, reaching with auditory rhythmic cues during motor rehabilitation might

PAGE 81

81 optimize the reaching task and result in long term improvements in adaptive variability of UE. The pattern of percent contribution of UE joints to total ApEn was not significantly different in the comfortable pace reaching condition between healthy controls and individuals with stroke. Specifically, in this reach to point task, PIP (index finger) joint provided the greatest contribut ion to total ApEn, followed by wrist, elbow and shoulder joints in a decreasing order. In contrast, in a previous study, in which participants reached to grasp a can with the more affected UE, we discovered that individuals with stroke exhibited an altered pattern of percent contribution of UE joints to total ApEn (Figure 4 6 ). In particular, wrist and elbow contributed significantly greater than PIP (index finger) joint. A major difference between the two studies is that in the previous study participants performed reach to grasp movements while here they did reach to point movements. The complex nature of reach to grasp might be responsible for this altered pattern of distribution of UE joints ApEn to total ApEn. Grasping a can requires greater involvement of distal UE joints (PIP) for manipulating the fingers around the can than reach to point task (Van Thiel, Meulenbroek, Smeets, & Hulst ijn, 2002) For individuals who have moderate UE deficits post stroke, such as our participants who had some, but limited, ability to voluntarily extend the fingers, manipulating the PIP joint of index finger around the can could potentially require too m uch effort resulting in a compensation strategy exhibiting greater use of wrist and elbow joints. Our preliminary findings show initial support that movement speed as well as auditory rhythmic cuing might act as appropriate task constraints to transition stereotypical movement patterns to more adaptable ones post stroke. Interestingly,

PAGE 82

82 further analysis revealed that there was no significant difference in corresponding ApEn estimates of shoulder, elbow, wrist and PIP (index finger) joints between comfortabl e reaching condition in healthy controls and metronome condition in individuals with stroke (Figure 4 7 ). Reaching with rhythmic auditory stimulation could facilitate a more normal functioning of the UE motor system so that UE adaptive variability in indiv iduals with stroke increased towards that of healthy controls. Gains in adaptive variability indicated enhanced complexity in the functioning of the motor system post stroke This study provides a strong rationale to incorporate task constraints such as, au ditory rhythm cues or UE training at faster speed in enhancing UE adaptive variability post stroke. However, careful and systematic manipulations of these task constraints are further required to ascertain the critical speed and frequency of auditory rhyth mic cues which cause the transition to more adaptable reach to point movements post stroke. Recently, Malcolm, Massie, and Thaut (2009) also demonstrated improvements in UE kinematics post stroke with an intervention based upon two week program of rhythmic auditory stimulation. Future studies should also investigate the long term benefits of auditory rhythm cues in enhancing UE adaptive variability post stroke. Future studies should also be conducted to understand the effect of movement speed and rhythm on the dynamics of inter joint coordination during different reaching conditions. There are certain limitations of this study. We did not control for the randomization of three reaching conditions. We believed that reaching faster or with a metronome first might influence the comfortable reaching condition, thus the comfortable pace condition was always performed first. Given the heterogeneity observed in stroke, this sample size was relatively small. However, moderate to large

PAGE 83

83 effect sizes were evident in b oth controls as well participants with stroke. Nevertheless large samples should be studied in order to investigate the severity influences of stroke on adaptive variability. Implications for R ehabilitation This study implies that reaching faster as well as external auditory rhythmic cuing might enhance the adaptive variability of UE in individuals with stroke. Adaptive variability is significantly impaired in shoulder, elbow, wrist and PIP (index finger) joints post stroke (Sethi, et al., 2009). Although therapy goals are not explicitly stated to enhance adaptive variability, an implied expectation of therapists is to achieve functional movements that are adaptable to successfully meet the demands of everyday changing tasks of our clients. To achieve, suc h adaptable functional movements, our clients require adequate adaptive variability. This study provides empirical evidence of ways to enhance UE adaptive variability post stroke. Specifically, UE rehabilitation interventions utilizing the principles of re aching faster or rhythmic entrainment might enhance UE adaptive variability post stroke.

PAGE 84

84 Table 4 1 Participant demographics: individuals with stroke Participant Gender Age Affected Side UE_FM Lesion Location Months after CVA 1 F 48 L 35 Right basal ga nglia 7 2 M 76 L 41 Right middle cerebral artery 102 3 M 62 L 27 Right middle cerebral artery 118 4 F 64 L 31 Right lacunar infarct 67 5 F 62 L 27 Posterior periventricular white matter 19 6 M 72 R 38 Left lacunar infarct 24 7 M 76 R 53 Left middle c erebral artery 174 8 M 77 R 38 Left pontine infarct 34 9 M 55 L 41 Right medial medullary infarct 43 10 M 72 L 30 Right middle cerebral artery 48 11 M 66 L 43 Right striatocapsular infarct 105 12 M 70 L 45 Right Cerebellar Stroke 14 13 M 68 R 31 Le ft middle cerebral artery 16 Mean ( SD) 67 (8.9) 36.9 (7.9) 53.3 (50.9)

PAGE 85

85 Table 4 2. Participant demographics: healthy controls Participant Gender Age Dominant Hand 1 F 61 R 2 F 43 R 3 F 51 R 4 M 62 L 5 F 62 R 6 F 56 R 7 F 58 R 8 Mean ( SD) F 57 56.25 ( 6.49) R Figure 4 1. Upper extremity marker set

PAGE 86

86 Figure 4 2. ApEn in various UE joints in healthy participants in three reaching conditions. (ApEn = Approximate Entropy ) Figure 4 3. ApEn in various UE joints in indivi duals with stroke in three reaching conditions (ApEn = Approximate Entropy; = significant )

PAGE 87

87 Figure 4 4. ApEn percent of various UE joints to total ApEn in healthy participants in three reaching conditions. (ApEn = Approximate Entropy) Figure 4 5. ApEn percent of various UE joints to total ApEn in individuals with stroke in three reaching conditions. (ApEn = Approximate Entropy)

PAGE 88

88 Figure 4 6. Approximate entropy (ApEn) percent of each joint to total ApEn during a reach to grasp task in healthy controls and individuals with stroke (* = significant) Figure 4 7 ApEn of various UE joints in healthy participants in comfortable p ace reaching condition compared to ApEn in individuals with stroke in rhythmic re aching conditions. (ApEn = Approximate Entropy)

PAGE 89

89 CHAPTER 5 DOES INTENSE FUNCTIO NAL TASK TRAINING EN HANCE U PPER E XTREMITY ADAPTIVE VARIABILITY POST STROKE? B ackground Stroke is a serious and disabling condition affecting over 795,000 individuals are affected by stroke in United States every year (Ame r ican Heart Association, 2010). Hemiparesis is one of the most common motor impairments post stroke (Barnes, Dobkin, & Bogousslavsky, 2005). Up to 85% of individuals with stroke exhibit upper extremity (UE) paresis immediately post stroke (Olsen, 1990). Mo st UE stroke rehabilitation interventions aim to improve motor control and enhance involvement of paretic UE in functional activities with intent to promote functional independence (Krakauer, 2006). UE movements post stroke are often stereotypical (Cirstea & Levin, 2000) and characterize d by low complexity in shoulder an d elbow flexion as measured by approximate e ntropy (ApEn) (Sethi,et al., 2009). This is problematic for functional use of the UE because an optimal amount of complexity provides individuals with a repertoire of movement st rategies to adapt movement patterns and successfully meet the demands of everyday changing tasks. For example, reaching for a soda pop can on a low surface would employ different movement strategies than reaching overhead for the same can. It might then se em intuitive to refer Loss of UE complexity or adaptive variability post stroke makes the motor system more rigid limits flexibility and might limit function. Stergiou, Harbourne and Cavanaugh (2006 ) indicated that the goal of neurological rehabilitation should be to facilitate the adaptive variability. Constraint induced movement therapy (CIMT) is one of the most widely studied UE rehabilitation

PAGE 90

90 interventions post stroke (Langhorne, et al., 2009) CIMT is a conglomeration of various intervention principles aimed to enhance the use and function of the more affected UE in individua ls with adult stroke (Taub, et al., 1993) ; (van der Lee, et al., 1999) ; (Dromerick, Edwards, & Hahn, 2000) CIMT consists of repetitive task oriented training, adherence enhancing behavioral strategies and constraining the use o f the more affected use of UE (Morris, Taub, & Mark, 2006) The typical CIMT protocol constitutes the engagement of more affected UE in functional task acti vities utilizing the principle of and task practice for 6 hours per day for 2 weeks. Shaping is a technique where individuals complete a functional task in successive approximations (Morris, et al., 2006) The use of the less affected hand is limited by constraining it in a mitt for 90% of the waking hours for two weeks. Furthermore, to ensure compliance with the therapy behavioral contracts are als o created with the person and his or her significant other. Intense repetitive functional task practice constitutes the hallmark of CIMT intervention. Systematic reviews conducted by Langhorne, et al., (2009) and Bonaiuti, et al., (2007) suggest the hig hest level of evidence for CIMT in UE stroke rehabilitation. The findings of the first multicenter EXCITE randomized control trial demonstrated positive motor gains post CIMT (Wolf, et al., 2006) Individuals with stroke not only were able to complete functional tasks faster but also improved the use of the paretic arm in common functional tasks. Significant gains in UE function were also seen with at least three hours of modified CIMT (Sterr, et al., 2002) A few studies have also shown kinematic changes post CIMT (Caimmi, et al., 2008; Wu, et al., 2007) Wu, et al., (2007) demonstrated that post CIMT, participants

PAGE 91

91 had significant improvements in movement duration produced smoother and straighter movement trajectories with more affected UE as compared to those who received conventional therapy. Cammini, et al., (2008) also observed favorable changes in paretic UE kinematics in indi viduals with stroke post CIMT. However, the effect of intense functional task practice such as, CIMT on UE adaptive variability using ApEn has not been studied. Optimal adaptive variability is a characteristic feature of a healthy well functioning motor sy stem and imparts adaptability to movements (Harbourne & Stergiou, 2009) Therefore, the primary purpose of this study was to investigate the effect of CIMT upon UE adaptive variability post stroke. We hypothesized t hat individuals with stroke would demonstrate sig nificantly greater ApEn in shoulder elbow and wrist flexion/extension joints angles in reach to grasp movements post CIMT. One of the preliminary steps before investigating the effect of CIMT upon UE adaptive variability is to establish the stability of ApEn over time using reliability analysis so that any changes seen in ApEn could be attributed to the intervention The reliability of ApEn could be determined from measurements of the same participants on two occasions: retest reliability of ApEn. Hence, the secondary purpose of this paper was to investigate the test retest reliability of ApEn in shoulder, elbow, and wrist flexion/extension joints angles in reach to grasp task post stroke. We hypothesized th at ApEn in shoulder, elbow and wrist flexion/extension joints angles would be reliable in reach to grasp movements in individuals with stroke. In order to identify a meaningful change in ApEn post CIMT minimal clinical important difference was also computed. The information gained from this study could provide preliminary evidence on whether CIMT enhances UE adaptive variability post stroke.

PAGE 92

92 Methods Research Design for C onstraint I nduced M otor T reatment P rotocol This study employed a one group pre test post t est design. Participants For reliability analyses 14 individuals with chronic stroke were recruited from a larger randomized controlled trial with a mean age of 66.07 years (8.03). Table 5 1 provides the demographic details and the Fugl Meyer UE subscal e (FM_UE) scores of the participants with stroke in reliability analysis. For the intervention study, 6 individuals with chronic stroke underwent CIMT with a mean age of 67 years (1 0.69 ). Table 5 2 provides the demogra phic details and the FM_ UE subscale s cores of the participants with stroke involved in the CIMT intervention University of Florida Institutional Review Board and North Florida/South Georgia Veteran Health Systems Research and Development Committee approved the protocol of this study. Partici pants were included if they: (1) were between the ages of 18 90 years of age; (2) had a single ischemic stroke at least 6 months prior; (3) were able to follow two step commands; (4) had no history of more than minor head trauma, subarachnoid hemorrhage, d ementia or other neural disorder/dysfunction, drug or alcohol abuse, schizophrenia, serious medical illness, or refractory depression; (4) were able to elevate UE in scapular plane (combination of flexion and abduction) at least 30 0 with at least 45 0 activ e elbow extension available during this movement; 5) were able to extend the wrist 20 0 and 2 fingers and the thumb 10 0 three times in a minute. Participants were excluded if they exhibited: (1) no active movement in UE; (2) spasticity greater than 2 on th e Modified Ashworth Scale (Pandyan, et al., 1999) ; (3) scores > 3 on Motor Activity Log (Blanton &

PAGE 93

93 Wolf, 1999) indicating poor use of UE; (4) ata xia, major sensory deficits, or hemi inattention/neglect. Procedures U pper extremity (UE) k inematic analysis of reach to grasp All participants underwent baseline UE kinematic testing prior to intervention. Fourteen participants recruited for the reliabili ty analysis were tested again between 24 hours up to one week apart. Of the 14 participants recruited for the reliability analysis, 4 participated in the two weeks of CIMT intervention and also underwent immediate post testing. In addition, 2 more particip ants participated in the two weeks of CIMT intervention and also underwent immediate post testing. Hence, 6 participants underwent CIMT intervention. Participants reached forward and grasped a soda can (56 mm in diameter; 208 mm circumference) first wit h the non paretic UE and then w ith the paretic UE at the Human Motor Performance Laboratory in the Brain Rehabilitation Research Center, at Veteran Affairs Medical Center Gainesville, Florida. Forty three reflective markers were secured to the various lan dmarks of the upper body. Marker placements were determined using the Plug In UE marker set defined by our laboratory. P articipants were seated on an adjustable, backless bench with knees bent at 90 of flexion and feet flat on the floor, hands palm down on the table in front of them and supported in 90 of elbow flexion by arm rests flush with the table. A filled soda can was placed at 80% middle finger) on a table directly in front of the affected shoulder of the participant. Healthy individuals use UE joints alone to reach for objects within this workspace; to

PAGE 94

94 obtain objects beyond this boundary; they typically involve anterior flexion of the trunk (Mark et al., 1997). Part icipants reached for the can, lifted it off the table, and put it back down as fast as they could and returned to the starting position. Participants performed four trials with the firs t serving as a practice trial. command. Participants performed discrete trials because majority of the functional tasks recorded using a 12 camera VICON motion capture system ( Vicon 612, Oxford Metrics, Oxford, UK) at a sampling frequency of 100 Hz. A preliminary frequency analysis of the reach to point data across all the conditions indicated that the range of signal frequencies that contain 99.99% of the overall signal power is between 1 and 10 Hz. Therefore, the sampling frequency was set at 100 Hz in order to be above a factor of ten higher than the highest frequency that might contain relevant signal. M idway through the study, the motion capture software was changed from Vicon Workstation to Vicon Nexus 1.3 (Oxford, UK). This system allowed for higher efficiency, better resolution and data within this system were sampled at 200 Hz. Eight of the 14 participants in the reliability analysis and two out of 6 in the CIMT intervention study, underwent testing using Vicon Workstation. The remaining participants were tested using Vicon Nexus. Hence, t he data collected using VICON Nexus were down sampled from 200 to 100Hz for even comparisons. Data processing Data analysis was performed on the last three trials. The data were the 3 D positional coordinates of each marker with respect to a laboratory coordinate system throughout the movement series. The data was then manually labeled, and reconstructed. We analyzed the unfiltered data in order to retain the inherent s tructure

PAGE 95

95 of variability of the kinematic data (Rapp, et al., 1993) The data co llected using VICON Nexus was down sampled from 200 to 100Hz for even comparisons. The entire movement cycle was analyzed. The start of reach was identified as the time point at which the velocity of the index finger marker exceeded 5% peak velocity and th e termination of reach as the time point at which velocity of this marker fell below 5% peak velocity. One degree of freedom in the sagittal plane (flexion/extension) was used to determine shoulder, elbow, and wrist joint angle. The kinematic data were the n modeled using SIMM (4.2, Santa Rosa, CA), which provided the shoulder, elbow, wrist and PIP (index finger) joints of angles Surrogate analysis A surrogation procedure was applied prior to computing ApEn utilizing Theiler et erify whether the kinematic data were deterministic (non phase randomization technique which removes the deterministic structure from the original shoulder, elbow, an d wrist joint angle time series creating 20 surrogate time series of each trial with the same mean, variance, and power spectrum as the original time series. ApEn was then computed on the original as well as each of the 20 surrogate time series. Significan t differences in ApEn between the original and 19 out of 20 surrogate time series would confirm the deterministic nature of the original data Approximate e ntropy After verifying th at the kinematic data was not random ApEn was obtained using the using MA TLAB code (R2009a, Natick, MA) developed by Kaplan and Staffin (1996) utilizing the algorithm provided by (Pincus, et al., 1991) ApEn is a meas ure of complexity and determines the randomness in a time series. S houlder, elbow, and wrist

PAGE 96

96 flexion/extension angle time series of all three trials were utilized for analysis of ApEn. Each joint angle time series was analyzed from the start of the reach t hrough the entire length of the respective time series including the pauses between the three trials. This procedure is not the same as computing ApEn on three individual trials of reach to grasp separately. In fact, the pauses between the three trials wer e evaluated as part of the temporal structure of the entire time series. Such an approach was adopted because ApEn is effectively a probability measure, which identifies whether small patterns of the time series repeat later in the entire time series. Thes e small patterns might not be repeated in a single trial of reach to grasp movement. This rationale justifies our approach to evaluate the temporal structure of the entire joint angle time series across all trials Hence, three time series were obtained: o ne for each of the shoulder, elbow and wrist joint angles. The most common method employed in the computation of ApEn is to identify repeating short patterns of length m across the entire shoulder joint angle time series. Starting with a vector of length m at point pi in the shoulder joint angle time series, the procedure involved counting the number of other vectors at other points pj (j i) in the same time series which have a similar pattern within r times the standard deviation of the shoulder joint angle time series. As a result, Cm(r), a count of the recurrence of vectors of length m was obtained. This same procedure was then repeate d for all vectors of length m in the shoulder angle time series and summing the logarithm of the results r is a similarity criterion, and provides the limits for assessing the nearness of adjacent data points in the shoulder joint angle time series. Ano ther parameter, lag, identified the number of time steps between points in one of the length m vectors. Biomechanical data analysis conventionally utilizes r =

PAGE 97

97 0.2 times the standard deviation of the time series, lag =1 and m = 2 (Slifkin & Newell, 1999) Thereafter, the log of this similarity count Cm(r), was normalized by the number of points in the shoulder angle time series. Following this, the recurrence of vectors of length m + 1 was obtained [Cm+1(r)], in the entire shoulder joint angle time series ApEn of shou lder angle was thus computed as the natural logarithm of the ratio of Cm(r) and Cm+1(r), as follows: ApEn (X m.r) = log [C m (r)/C m+1 (r)] (5 1) ApEn of elbow and wrist joint angles were also computed in the same manner. Because the length of the data could affect ApEn values, we normalized the ApEn values of shoulder, elbow and wrist of each participant to their data length and then multiplied the ratio with a constant equal to 100. A more detailed description of the computation of ApEn can be reviewe d in the Appendix of Slifkin and Newell (1999). In general, a vector of shorter length repeats more often than a longer one within a time series, thus the lowest possible ApEn value can be the natural logarithm of 1, which is 0, and negative values cannot be obtained. ApEn values range from 0 to 2. In a highly periodic or regular time series, values of Cm(r) can be similar to Cm+1(r) and ApEn = 0 Hence, smaller values characterize a more regular time series where similar patterns are more likely to follow one another. In contrast, high ApEn values, suggest a highly irregular time series, where the predictability of subsequent patterns is low and ApEn could be close to 2 (Stergiou et al. 2004). Our previous findings showed that individuals with stroke h ave significantly low er ApEn shoulder, elbow and wrist estimates for reach to grasp task as compared to healthy controls (Sethi et al, 2009), suggesting low UE adaptive variability post stroke.

PAGE 98

98 In this study we aimed to examine whether CIMT based intense f unctional task practice could significantly enhance ApEn of shoulder, elbow and wrist joints depicting higher UE adaptive variability post stroke C onstraint i nduced m ovement t reatment All participants underwent a 2 week program of intense functional tas k practice based upon CIMT. During the 2 week program, subjects wore a mitt on their unaffected UE for 90% of their waking hours. The mitt allowed gross UE movements, but prevented object manipulation. The mitt was removed overnight while sleeping, bathing and for contracted activities for safety or need. All participants attended 4 hour therapy sessions for therapist guided functional task practice with the affected extremity for 5 weekdays for two weeks. During this therapy participants performed functi onal activities, where each activity was practiced for each activity for 20 minutes. The therapist graded the activities throughout the therapy program by changing the size of the objects manipulated (e.g., smaller objects required fine motor control, and large objects were more difficult for individuals to grasp who had poor finger extension and weak grasp), the speed of movements required by the task (e.g. faster was more difficult), the weight of the object manipulated (e.g., heavier objects were more di fficult), and the location in space related to the body that these activities took place (e.g., generally, the higher the UE elevation with elbow extension the greater the difficulty). P articipants were given three tasks to practice on the weekends and fil led out a log recording the activities performed with their paretic UE when not in therapy and when they removed the mitt

PAGE 99

99 Statistical Analysis Dependent one tailed t tests were conducted to compare ApEn of shoulder, elbow and wrist joints values between the original and surrogate time series to confirm the deterministic nature of the original data. After verifying determinism in the joint angle series reliability tests were conducted. Relative reliability was conducted using ICC and Bland and Altman plot s revealed the absolute reliability of ApEn across all joints between the two time sessions Reliability A nalyses Relative reliability Relative reliability of ApEn of shoulder, elbow and wrist joints was estimated using intraclass correlation coefficient (ICC) (2, 1 ) (Portney & Watkins, 2000). A repeated measures analysis of variance (ANOVA) with test session, as an independent variable was used to separate total variance into: variance due to test sessions, variance due to differences in participants, and variance due to error. If EMS represents the error mean square, BMS represents the variance between test sessions and RMS represents the between session mean square and n denotes the number of participants then: ICC (2,2) = BMS EMS/BMS + ((RMS EMS)/n ) (Por tney & Watkins, 2000) (5 2 ) ICC accounted for relative reliability between two or more repeated measures on ApEn of shoulder, elbow and wrist joints ICC ranges from 0.00 1.00 where an ICC above .75 is indicative of excellent reliability (Fleiss, 1986). Absolute reliability Bland Altman plots were constructed by plotting the test retest mean of ApEn shoulder on the x axis versus between session differences of ApEn shoulder on the y axis (Bland & Altman, 1986) The data from these plots was further examined for its

PAGE 100

100 magnitude, range and distribution aroun d the zero line. Bland and Altman analyses involved the following calculations: = mean difference of ApEn shoulder between 2 test sessions (test session 2 minus test session1 SD diff = standard deviation of the differences between 2 te sts sessions SE (standard error) of = SD diff 3 ) 95% confidence intervals of (95% CI) = t SE (5 4 ) where the value of t will be obtained fr om the t table with 13(n 1) degrees of freedom (Bl and & Altman, 1986) Standard error of measure (SEM) and minimal detectable change (MDC) was also calculated for ApEn shoulder. The SEM represents the within subject variability. SEM was computed by taking the square root of the mean square residual error obtained from the ANOVA table used in ICC calculation ( Lexell & Downham, 2005 ; Portney & Watkins, 2000). SEM percentage was computed as: SEM% = (SEM/mean) 100 (5 5 ) where mean is the mean of all observations of ApEn shou lder from test session 1 and 2. SEM% provides the limit for the smallest change indicating a true improvement for a group of participants (Lexell & Downham, 2005; Portney & Watkins, 2000). Finally, the MDC was calculated using the formula: MDC 95 = SEM 1.96 (5 6 ) where 1.96 is the 2 variance of 2 measurements. MDC indicates the magnitude of change necessary to

PAGE 101

101 exceed measurement error of two repeated measures at a specified CI (Lexell & Downham, 2005 ; Portney & Watkins, 2000). Similar to SEM, MDC percentage was also computed: MDC% = (MDC 95 /mean) *100 (5 7 ) where mean is the mean of all observations of ApE n shoulder from test session 1 and 2. MDC% provides the smallest change that indicates a real change in a single participant (Lexell & Downham, 2005) Similarly Bland Altman plots, SEM, SEM%, MDC 95 and MDC% were also computed for ApEn elbow and wrist joints. Systematic bias was investigated by conducting Wilcoxon Signed Rank tests to identify no significant differences between the sample means of test versus sample means of retest values of ApEn shoulder joint. The data were also evaluated for heteroscedasticity, which reflects an increase in th e amount of random error as the measured value increases. To investigate heteroscedasticity, the absolute difference between the test and retest values of ApEn shoulder was correlated with the means of the participants test and retest values of ApEn should er. A weak non significant correlation suggests lack of heteroscedasticity (Bland & Altman, 1986) Similarly, systematic bias and heteroscedasticity were also investigated for ApEn shoulder and wrist joints. Statistical Testing for Intervention E ffects Non parametric analyses were employed due to the sm all sample size using SPSS (17.0, Chicago, IL). Wilcoxon Signed Rank test were used to investigate differences for ApEn shoulder, and wrist joints before and after CIMT with a level of significance at p< 0.05 with Holm s step down correction procedure. Hol m s step down procedure involves sorting the significance values obtained for each hypothesis (here

PAGE 102

102 each joint) in an ascending order. Consequently, the hypothesis with the lowest significance value was evaluated at p = 0.016 (0.05/3) If the previous hypo thesis was significant then the next hypothesis was evaluated at p = 0.025 (0.05/2) and the last at p = 0.05. The analysis was not continued further if any hypothesis was rejected. Additional analysis was also conducted to examine whether CIMT could faci litate improve ment in ApEn at shoulder, elbow and wrist joints in individuals with stroke to healthy participants. In a previous study (Sethi et al., 2009), our sample of healthy individuals ApEn values of shoulder (0.14), elbow (0.16) and wrist (0.24) joi nts during a reach to grasp task. ApEn values obtained at shoulder, elbow and wrist joints prior to and post CIMT were expressed as a percentage of the ApEn values at the respective joints ob tained by healthy participants. Specifically, the ratio of indivi dual ApEn shoulder values of each participant (pre and post CIMT) to the healthy ApEn shoulder value was multiplied by 100 and expressed as percentage. Similarly, the ratio of the mean of the ApEn shoulder value of all six participants was also multiplied by 100 and expressed as percentage. Likewise, percentages were also obtained of individual and mean ApEn elbow and wrist values pre and post CIMT with respect to respective normative values Results Testing Determinism in J oint A ngles Dependent one tail ed t tests between the original and surrogate time series to confirm the deterministic nature of the original data revealed significantly greater shoulder, elbow, wrist and PIP (index finger) joints ApEn in surrogate time series for both controls (p<0.05) and stroke (p<0.05) groups. These results suggested that original shoulder, elbow, wrist and PIP (index finger) joint angle time series were deterministic in nature and not derived randomly.

PAGE 103

103 Reliability A nalyses Relative reliability ICC ranges from 0.00 t o 1.00 where an ICC between 0.6 0.7 suggests good reliability and an ICC above .75 is indicative of excellent reliability (Fleiss, 1986). Based upon these criteria, relative reliability was found to be excellent for ApEn shoulder (0.81) and elbow (0.84) an d good for ApEn wrist (0.70) (Table 5 3) Absolute reliability The residuals errors between test and retest values of ApEn across all joints were normally distributed. Table 5 3 depicts the 95% CI of of ApEn shoulder, elbow and wri st joints. Bland and Altman statistics demonstrated no systematic bias in ApEn shoulder, elbow and wrist joints, where zero was included in the 95% CI of in ApEn of all joints between session 1 and session2 (Figure s 5 1, 5 2, and 5 3 ). Further, Wilcoxon Signed Rank tests revealed no significant bias between values of ApEn shoulder, elbow and wrist joints obtained at session 1 and 2 (p >0.05). Visual inspection of all Bland and Altman plots demonstrated lack of heteroscedasticity ( Figu res 5 1, 5 2, and 5 3 ). Weak and non significant (p >0.05) Spearman correlation between values of ApEn shoulder, elbow and wrist joints obtained at session 1 and 2 further confirmed lack of heteroscedasticity in the data. Standard error of measurement and minimal detectable change Lower SEM% values indicate lower error due to measurement than higher SEM% ( Lexell & Downham, 2005 ). Table 5 3 shows the SEM% of ApEn shoulder, elbow and wrist joints. ApEn shoulder and elbow obtained moderate SEM% (<35%) while ApEn wrist was larger (SEM ~ 39%) Lower MDC% suggests that smaller changes in the outcome measure are sufficient to detect a real change to treatment than higher

PAGE 104

104 MDC% (Beckerman, et al., 2001) Table 5 3 shows the MDC% of ApEn shoulder, elbow and wrist joints. Similar to SEM%, ApEn wrist joint resulted in greater MDC% (~105%) than the other two joints. Effect of CIMT on UE A daptive V ariability Mean values of ApEn shoulder, elbow and wrist joints were greater post CIMT (Figure 5 4 ). Table 5 4 and 5 5 depict both individual percent change and means in ApEn shoulde r, elbow and wrist joints post CIMT respectively Mean percent change in ApEn shoulder joint (~85%) exceeded MDC%. This was true for 4 of the 6 participants. Similarly, mean percent change in ApEn wrist joint (~114%) exceeded MDC%. However, only 3 of the 6 participants actually obtained ApEn values larger than MDC%. The mean percent change in ApEn elbow joint (~63%) did not exceed MDC% and only 2 of the 6 parti cipants exceeded MDC%. Table 5 6 also shows normalized pre and post ApEn values at shoulder, elbo w and wrist joints in participants with stroke as a percentage of ApEn values of respective joints of healthy participants. M ean ApEn estimates at shoulder, elbow, and wrist joints post CIMT did not approximate the norm ative values of ApEn Wilcoxon Signed Rank test revealed that there was no significant difference between ApEn shoulder pre CIMT (Mdn =0.02) as compared to post CIMT (Mdn =0.06), T=0, p = .02, r = 1.05 Similarly, Wilcoxon Signed Rank test further demonstrated no significant difference in ApE n elbow pre CIMT (Mdn =0.05) as compared to post CIMT (Mdn =0.09), T=1 p = .04, r = .64. Likewise, change in ApEn wrist also did not achieve statistical significance (pre CIMT: Mdn =0.09 ; Post CIMT: Mdn =0.20), T=3 p = .14, r = .65 T he lowest p value obtained at ApEn shoulder joint was not statistically significant at the corrected level of significance (p < 0.016). Therefore, based upon Holm s step

PAGE 105

105 down procedure further analyses between ApEn elbow and wrist joints also did not reveal statistical sign ificance. Discussion The goal of this study was to examine whether changes in approximate entropy were promoted by therapy based on CIMT. However, in order to adequately test this result, we needed to determine the reproducibility and MDC% of adaptive va riability of shoulder, elbow and wrist joints post stroke as measured by ApEn. In our sample the relative test retest reliability utilizing ICC was good for ApEn wrist and excellent for ApEn shoulder and elbow. Bland and Altman plots revealed stable perf ormance with minimal systematic variance in ApEn shoulder, elbow and wrist joints between test session 1 and 2. Further, SEM and SEM% was utilized to evaluate the measurement error of ApEn across the three UE joints. The SEM and SEM% accounts for the small est change required to depict real improvement across participants (Lexell & Downham, 2005) In the present study SEM% ranged from 25% to 39%. In particular, ApEn shoulder and elbow obtained less than moderate SEM% (<35%), whereas SEM% of ApEn wrist was 39%. Based on these data, the results sugg est that more than moderate (measurement error of 35%) changes were required to indicate a real change in adaptive variability at various UE joints post stroke. Based upon the SEM values MDC and MDC% was also obtained. MDC and MDC% indicates the magnitude of change necessary to exceed measurement error of two repeated measures at a 95% confidence interval (Lexell & Downham, 2005) The MDC% estimate for ApEn resulted in 70% (shoulder), 78% (elbow) and 105% (wrist). Specifically, the MDC% estimate of ApEn wrist was of a greater magnitude as compare d to ApEn shoulder and elbow. Based upon the current sample of stroke

PAGE 106

106 survivors, these results indicate that a larger magnitude of change would be required in ApEn wrist than ApEn shoulder and elbow to indicate a real improvement in adaptive variability po st stroke. This was the first study, to our knowledge, to examine the test retest reliability and MDC% of UE adaptive variability post stroke. Having determined that ApEn measures are reliable across sessions and calculating the MDC%, an appropriate de termination of ApEn treatment response could be made. Although adaptive variability of shoulder, elbow and wrist joints increased post CIMT, statistical significance was not achieved. Similar changes were observed in clinical scales, such as Wolf motor fun ction test and FM_UE (Figures 5 5 and 5 6) Specifically, mean FM_UE increased and mean WMFT decreased post CIMT suggesting gains in motor control and performance respectively. The lack of statistical significance is most likely because of the lack of powe r due to small sample size. M oderate to high effect sizes were obtained in ApEn across all joints bef ore and after CIMT. In particular ApEn wrist and elbow joints resulted in moderate effect size (0.65) and ApEn shoulder joint resulted in large effect siz e (1.05) post CIMT. However P ost hoc power analysis revealed, only 12% power was achieved for ApEn wrist joint, 30% for ApEn elbow joint and 68% for ApEn shoulder joint. Our sample size of six participants was not sufficient to obtain statistical signific ance. Furthermore, mean ApEn estimates at shoulder, elbow, and wrist joints post CIMT did not approximate the norm ative values of ApEn (Table 5 6 ). T hough the positive change in ApEn post intervention brought individuals with stroke closer to the healthy c omplex state there was still room for further improvement. However inspection of individual participant data

PAGE 107

107 revealed that participant 2 with a left middle cerebral infarct alone obtained greater ApEn percentage across all joints with respect to healthy participants. Looking at the mean and individual percent change of ApEn across shoulder, elbow and wrist joints indicated a majority of the participants exceeded the MDC% estimates in ApEn shoulder and wrist joints However, this was not the case at the elbow joint (Tables 5 4 and 5 5) Other studies have shown that motor control at the elbow after stroke such as elbow range of motion is more resistant to treatment than at the other joints. Caimmi et al., (2008) observed persistent reduced elbow extensi on in eight individuals with stroke following CIMT. Similar findings were demonstrated in a randomized clinical trial comparing the effect of functional task practice against strengthening of hemiparetic arm in 14 indiv iduals with stroke (Corti McGuirk, W u, & Patten, 2010). Functional task practice did not result in gains in e lbow extension. A daptive variability at shoulder and wrist joints might be more susceptible to change than at elbow joint post CIMT in individuals with stroke. Redundancy in the cort icospinal tracts (CST) with ipsilateral innervations of anterior CST might provide additional motor inputs to proximal joints, resulting in greater gains in ApEn shoulder (Colebatch & Gandevia, 1989) However, this argument cannot be made for the wrist joint, a nd at this point the specific neural and biomechanical mechanisms responsible in augmenting UE adaptive variability post CIMT are not certain. Examining the change in adaptive variability at an individual level revealed interesting findings. P articipant 5 with a stroke in the posterior limb of interior capsule did not exceed MDC% at any of the joints. Further examination revealed that participant 5 only increased ApEn at shoulder 29%, and 7% at the elbow joint. In contrast,

PAGE 108

108 participant 5 declined in ApEn at wrist joint by 40% post CIMT. S troke in the posterior limb of interior capsule ha s been associated with poor prognosis (Shelton & Reding, 2001) However, participant 5 improved on FM_UE (Table 5 7) Possibly, FM_UE and ApEn measure different constructs of motor control. In particular, most of the items perf ormed in FM_UE are static, where participants are asked to perform specific movement patterns. Any deviations in these movement patterns represent impairment in UE motor control and results in low score on the test. In contrast, the reach to grasp task ana lyzed in this study was dynamic and did not demand specific movement patterns across joints. Participants were only instructed to grasp the can from the side kept at a specific location on the table. Measures of adaptive variability, such as ApEn essential ly quantify the complexity in movement. Greater movement complexity indicates that an individual has the ability to utilize mu ltiple options to grasp the can M ovement complexity while grasping a can is different than grasping a can in a specific manner as tested in FM_UE. Hence, gains in FM_UE might not necessarily be associated with g ains in adaptive variability. A priori power analysis utilizing these results indicated that at least 28 participants with stroke would have been required to obtain a sign ificant effect (p < 0.01) of adaptive variability of shoulder and wrist joints post CIMT. Future studies with larger sample size need to investigate the effect of CIMT upon UE adaptive variability. Future studies also need to be conducted to understand the link between UE adaptive variability and neural reorganization in the brain post CIMT. Furthermore, studies also need to focus on the relationship between musculoskeletal factors such as muscle strength, and inter joint coordination and UE adaptive variab ility.

PAGE 109

109 There were certain limitations of this study. First, a reliability estimate utilizing a large sample size might result in lower MDC% estimates. However, these data are important, as this was the first attempt to identify the reproducibility of ApE n in reach to grasp movements post stroke. Second, the results of the current reliability analysis could not be generalized to other UE movement tasks. To our knowledge, this was the first study conducted to investigate the effect of one of the most comm only studied UE stroke rehabilitation interventions post stroke, CIMT, upon UE adaptive variability. The positive change in mean adaptive variability across all joints post CIMT provides preliminary support to our hypothesis, that CIMT might enhance UE ada ptive variability. Gains in adaptive variability post CIMT indicate d enhanced complexity in the func tioning of the motor system in individuals with stroke. A daptive variability provides individuals with an abundant repertoire of movement strategies to ada pt movement patterns and successfully meet the demands of everyday changing tasks. For instance: we employ different type of contractions of muscles and range of motion to pick an object from a level surface against overhead reaching of the same object. In addition, adaptive variability also imparts adaptability to modify movement in the event of perturbations. For instance: we modify our grasp patterns depending upon the size and shape of the objects. Furthermore, while holding a wet glass filled with wate r, our grasp becomes stronger as soon as we realize that the object might slip in our hands to prevent a spill. Hence, adaptive variability not only offers multiple options for task performance but also provides adaptability in movement. However, individua ls with stroke often exhibit stereotypical movement patterns with limited repertoire of movement strategies to accomplish daily living tasks. Moreover,

PAGE 110

110 adaptability of UE movement is markedly reduced post stroke (Sethi et al., 2009). Hence, CIMT might augm ent UE adaptive variability post stroke.

PAGE 111

111 Table 5 1. Participant demographics: individuals with stroke for reliability analysis Participant Gender Age Affected Side UE_FM Lesion Location Months after CVA 1 M 76 L 41 Right middle cerebral artery 102 2 M 62 L 46 Right M1, middle cerebral artery 48 3 F 70 L 44 Right Striatoscapular infarct 131 4 F 66 R 58 Left Midlle/Posterior cerebellar artery 102 5 M 73 R 45 Left medullary/brainstem infarct 103 6 M 76 R 53 Left middle cerebral artery 174 7 M 55 L 4 1 Right medial medullary infarct 43 8 M 66 L 43 Right Striatoscapular infarct 105 9 F 47 L 35 Right basal ganglia 7 10 M 62 L 27 Right middle cerebral artery 118 11 F 64 L 31 Right lacunar infarct 67 12 F 62 L 27 Posterior ventricular white matter 19 13 M 72 R 38 Left lacunar infarct 24 14 M 77 R 38 Left pontine infarct 34 Mean ( SD) 66.07 ( 8.03) 40.5 ( 8.8) 76.9 (49.4)

PAGE 112

112 Table 5 2. Participant demographics: individuals with stroke for intervention analysis Participant Gender Age Affected Side U E_FM Lesion Location Months Post CVA 1 M 76 L 41 Right Middle Cerebral Artery 102 2 F 49 R 24 Left Middle Cerebral Artery 15 3 F 64 L 31 Right Periventricular White Matter 67 4 F 62 L 27 Right Posterior periventricular white matter 19 5 F 78 L 40 Post erior Limb of Internal Capsule 24 6 M 61 R 29 Left Middle Cerebral Artery 24 Mean ( SD) 67 ( 10.69) 32 ( 6.9) 41.83 ( 35.01) Table 5 3. Reliability estimates ICC 95% Confidence Interval SEM SEM% MDC MDC% ApEn Shoulder .81 .01 .02 .008 25.31 .02 7 0.17 ApEn Elbow .84 .03 .05 .015 28.35 .04 78.60 ApEn Wrist .70 .07 .10 .032 37.81 .09 104.81 ApEn: Approximate entropy ; ICC: Intraclass correlation; SEM: Standard error of measurement; SEM%: Standard error of measurement percent ; MDC: Minimal detecta ble change; MDC%: Minimal detectable change percent.

PAGE 113

113 Table 5 4. Individual percent change of ApEn in various UE joints after CIMT Participant Change ApEn Shoulder (%) Change ApEn Elbow (%) Change ApEn Wrist (%) 1 96.74 51.94 28.84 2 110 105.14 103.77 3 76.09 53.50 373.86 4 68.60 15.38 213.32 5 29.90 7.16 39.91 6 241.35 296.39 425.61 ApEn: Approximate entropy; CIMT: Constraint induced movement therapy Table 5 5. Mean change of ApEn in various UE joinits after CIMT Pre CIMT (Mean SE) Post CIMT (Mean SE) Mean Change (%) ApEn Shoulder 0.04 (.01) 0.08 (.02) 84.63 ApEn Elbow 0.05 (.01) 0.08 (.02) 63.23 ApEn Wrist 0.09 (.04) 0.20 (.08) 113.91 ApEn: Approximate entropy; CIMT: Constraint induced movement therapy

PAGE 114

114 Table 5 6. Individual pre and post percent of ApEn in various UE joints in individuals with stroke normalized to healthy participants Participant ApEn Shoulder Pre (%) ApEn Shoulder Post (%) ApEn Elbow Pre (%) ApEn Elbow Post (%) ApEn Wrist Pre (%) ApEn Wrist Post ( %) 1 9.54 18.78 15.13 22.99 17.36 22.37 2 74.04 155.46 66.24 135.90 123.87 252.43 3 29.17 51.37 25.06 38.47 24.54 116.31 4 11.25 18.97 17.05 14.43 7.20 22.58 5 58.62 76.16 67.31 72.13 51.60 31 6 10.50 35.86 11.81 46.83 11.47 60.31 Mean 32.19 59.43 33.77 55.12 39.34 84.17 ApEn: approximate entropy Table 5 7. Individual pre and post CIMT FM_UE and WMFT scores in individuals with stroke Participant FM_UE Pre FM_UE Post FM_UE% Change WMFT Pre (sec) WMFT Post (sec) WMFT% Change 1 41 39 5 22 7 68 2 24 24 0 71 38 47 3 31 38 23 .4 4.6 13 4 27 35 30 22 5 78 5 40 48 20 3.4 3.3 2 6 29 34 17 21 3 85 FM_UE: Fugl Meyer Upper extremity subscale; WMFT: Wolf motor function test ; CIMT: Constraint induced movement therapy.

PAGE 115

115 Figure 5 1. Bland and Altman plot of ApEn shoulde r (ApEn: Approximate entropy ; dotted line indicates 95% confidence interval ) Figure 5 2. Bland and Altman plot of ApEn elbow (ApEn: Approximate entropy ; dotted line indicates 95% confi d ence interval )

PAGE 116

116 Figure 5 3. Bland and Altman plot of ApEn wrist ( ApEn: Approximate entropy ; dotted line indicates 95% confidence interval ) Figure 5 4. ApEn of UE joints post CIMT( ApEn: Approximate entropy ; CIMT : Constraint induced movement therapy )

PAGE 117

117 Figur e 5 5 WMFT scores post CIMT (WMFT: W olf motor function test scores ; CIMT : Constraint induced movement therapy ) Figure 5 6 FM_UE post CIMT (FM_UE: Fugl Meyer upper extr emity subscale scores ; CIMT : Constraint induced movement therapy )

PAGE 118

118 CHAPTER 6 CONCLUSION: INTEGRAT ING THE FINDINGS Upper extremity (UE) impairme nt is one of the most frequent impairments after stroke (Gresham, et al., 1975) The damage to the motor system caused by the stroke results in imperfect motor control, often exhibited as atypic al or stereotypical movement patterns. One hallmark of this dysfunctional motor system is the high variability present in several movement parameters, such as upper extremity joint range of motion, movement time, and peak velocity, when variability is conc eptualized from the traditional motor control perspective (Cirstea & Levin, 2000) Movement variability under these traditional motor control theories is considered as undesirable noise in the motor output (Stergio u, Buzzi, Kurz & Heidel, 2004), and therefore error. Hence, the goal of traditional neurorehabilitation is to curtail UE movement variability to enhance motor control and function (Bobath, 1990). However, contemporary motor control theories, such as dynami cal systems theory, consider variability as a characteristic of movement that indicates a healthy, well functioning motor system and a consequence of motor learning (Bernstein, 1967; Kamm, et al., 1990) Variability in movement also reveals the inherent co mplexity of the system components and their functional interactions (Vaillancourt & Newell, 2002) In general, variability in movement imparts adaptability to the motor system ( Harbourne & Stergiou, 2009) Therefore, we suggest that the functional aspects of variability could be referred as adaptive variability. Lipsitz and Goldberger (1992) suggested that the functional aspects of variability decline as a function of aging an d disease. Because of the damage to motor neural networks from stroke, it would then seem intuitive that individuals with stroke might also exhibit reduced adaptive variability in upper extremity movement. Therefore, the first

PAGE 119

119 study of this dissertation ai med at determining whether adaptive variability was altered in reach to grasp movements due to damage to the motor system caused by stroke. Specifically, we investigated whether adaptive variability in shoulder, elbow, wrist and proximal interphalangeal (P IP of index finger) flexion/extension joint angles d uring reach to grasp movements wa s reduced in individuals with stroke as compared to age matched healthy participants. UE adaptive variability was quantified by computing the approximate entropy (ApEn) of shoulder, elbow, wrist and PIP (index finger) flexion/extension joint angles. Our findings revealed that adaptive variability was significantly reduced across all joints post stroke as compared to healthy participants. These findings indicate that adaptab ility in reaching movements is drastically reduced post stroke. Clinically, these findings imply that individuals with stroke might not be able to modify reach to grasp movement in the event of perturbations or changes in the task conditions. For instance, individuals with stroke might have difficulty in adapting grasp patterns depending upon the size and shape of the objects (Raghavan, Santello, Gordon, & Krakauer, 2010) Further, individuals with stroke are also l imited in the repertoire of movement strategies employed to accomplish a task. Because of reduced adaptive variability after stroke, the dynamical systems perspective would suggest that a goal for therapy should be to increase adaptive variability in in dividuals with stroke to enhance UE motor control and function. Utilizing the principles of DST, adaptive variability might be enhanced by changing certain task constraints, known as control parameters (Newell, 1986). Task constraints are variables that mi ght be highly specific, such as myelination, particular muscle strength, movement speed or nonspecific, such as emotional or motivational aspects. Changing the

PAGE 120

120 appropriate task constraints might transition the stable or stereotypical movement patterns with low adaptive variability to more adaptable ones. The next logical step then was to identify the appropriate task constraints that might be potential control parameters. Movement speed has been shown to drive the motor system from one stable state (or move ment pattern) to another in UE utilizing hand movements (Kelso, 1984) and gait (Diedrich & Warren, 1995) In addition, reaches made with auditory rhythmic cues were also known to augment the UE motor control post stroke (Thaut, et al., 2002) Therefore, the second study of this dissertation aimed to determine whether UE adaptive variability could be enhanced immediately when reaching naturally at a faster speed and/or to auditory rhyt hmic cues in individuals with stroke. We hypothesized that individuals with stroke would exhibit significantly greater shoulder, elbow, wrist and PIP (index finger) flexion/extension joint angles ApEn while reaching at a faster speed and/or to auditory rhy thmic cues versus reaching at comfortable pace. Our results indicated that shoulder, elbow, and wrist joints adaptive variability significantly increased while reaching faster post stroke. Similarly, auditory rhythmic cues also significantly enhanced shou lder, elbow, wrist, and PIP joints adaptive variability duri ng reaching. This study provided empirical evidence of ways to enhance UE adaptive variability post stroke. Clinically, UE rehabilitation interventions utilizing the principles of reaching faster or rhythmic entrainment might enhance UE adaptive variability post stroke. The results of the second study demonstrated that it was possible to make reaching movements more adaptable post stroke within one session. In fact, adaptive variability made whil e reaching with auditory rhythmic cues in individuals with stroke

PAGE 121

121 approximated to the adaptive variability of healthy participants. However, these findings only indicated the immediate benefits of reaching faster and with auditory rhythmic cues upon UE ada ptive variability. The next step was to investigate whethe r motor rehabilitation, could augment UE adaptive variability and transition the motor system towards normal. We chose to study Constraint Induced Movement Therapy (CIMT) because it is the most stud ied intervention with clear, but limited efficacy post stroke (Langhorne, et al., 2009) Hence, the third study of this dissertation aimed to examine the effect of CIMT in enhancing the UE adaptive variability pos t stroke. We hypothesized that CIMT could enhance adaptive variability of shoulder, elbow and wrist joints. We also investigated test retest reliability of ApEn across shoulder, elbow and wrist joints utilizing a comprehensive battery of statistical tools. Our findings demonstrated that approximate entropy, as a measure of adaptive variability was reliable across the three joints. Further, UE adaptive variability showed gains post CIMT. However the results were not statistically significant, most likely due to our small sample s ize of only six participants. We suggest future work with larger sample size to investigate the effect of CIMT upon UE adaptive variability Overall, a daptive variability is a characteristic fe ature of healthy motor system. In partic ular, adaptive variability imparts adaptability to the motor system. Further, ApEn as a measure of adaptive variability provides a unique and different perspective of motor control than com monly used clinical measures used in UE stroke rehabilitation such as Fugl Meyer UE subscale (FM_UE) and Wolf Motor Function Test (WMFT) In particular, most of the items performed in FM_UE require participants to perform specific movement patterns. Any deviations in these movement patterns represent impairment

PAGE 122

122 in UE moto r control and results in a low score on the test. Successful completion of a task constraints were altered. For instance, successful grasp of a can from a low table ( as performed in WMFT) might not generalize to successful grasp of the same can from a higher surface. In contrast, m easures of adaptive variability, such as ApEn essentially quantify the complexity in movement. Greater movement complexity indicates that an individual has the ability to utilize mu ltiple options to grasp the can. Specifically, an individual with greater UE adaptive variability might be more adaptable to grasp the can from different surfaces. Hence, u nderstanding the adaptive variability in UE movements provided a unique approach to examine the health or functionality of the motor control system post stroke and offer ed additional ways to describe the impairments in motor control post stroke. Results of this dissertation suggest that adaptabilit y of UE movements is markedly reduced post stroke. Developing adaptive variability is indicative of the development of greater functionality in the motor system. Hence, one of the goals of neurological rehabilitation should be to facilitate the adaptive va riability. UE rehabilitation interventions utilizing the principles of reaching faster or rhythmic entrainment might enhance UE adaptive variability post stroke. In addition, one of the widely studied motor rehabilitation interventions such as CIMT might a lso seem to augment UE adaptive variability post stroke. Future studies should aim at investigating the neurological correlates of adaptive variability in UE movements. Future studies also need to investigate whether practicing with error could augment ada ptive variability in UE movements post stroke

PAGE 123

123 LIST OF REFERENCES Adams, J. A. (1971). A closed loop theory of motor learning. J Mot Behav, 3 (2), 111 149. Alexander, M. P. (1994). Stroke rehabilitation outcome. A potential use of pre dictive variables to establish levels of care. Stroke, 25 (1), 128 134. American Heart Association. (2010). Heart disease and stroke statistics. http://www.americanheart.org/downloadable/heart/1265665152970DS 3241%20HeartStrokeUpdate_2010.pdf Accessed July 31, 2010, Dallas, TX. Beckerman, H., Roebroeck, M. E., Lankhorst, G. J., Becher, J. G., Bezemer, P. D., & Verbeek, A. L. (2001). Smallest real difference, a link between reproducibility and responsiveness. Qual Life Res, 10 (7), 571 578. Begliomini, C., Nelini, C., Caria, A., Grodd, W., & Castiello, U. (2008). Cortical activations in humans grasp related areas depend on hand used and handedn ess. PLoS One, 3 (10), e3388. Bernstein, N. (1967). Coordination and regulation of movement. New York: Pergamon Press. Bland, J. M., & Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, 1 (8476), 307 310. Blanton, S., & Wolf, S. L. (1999). An application of upper extremity constraint induced movement therapy in a patient with sub acute stroke. Phys Ther, 79 (9), 847 853. Bobath, B. (1990). Adult hemiplegia: Evaluation and treatment (3rd ed .). Oxford UK: Butterworth Heinemann. Bonaiuti, D., Rebasti, L., & Sioli, P. (2007). The constraint induced movement therapy: a systematic review of randomized controlled trials on the adult stroke patients. Eura Medicophys, 43 (2), 139 146. Buzzi, U. H., & Ulrich, B. D. (2004). Dynamic stability of gait cycles as a function of speed and system constraints. Motor Control, 8 (3), 241 254. Caimmi, M., Carda, S., Giovanzana, C., Maini, E. S., Sabatini, A. M., Smania, N., et al. (2008). Using kine matic analysis to evaluate constraint induced movement therapy in chronic stroke patients. Neurorehabil Neural Repair, 22 (1), 31 39. Catalano, J. F., & Kleiner, B. M. (1984). Distant Transfer in Coincident Timing as a Function of Variability of Practice. Perceptual and Motor Skills, 58 (3), 851 856.

PAGE 124

124 Cirstea, M. C., & Levin, M. F. (2000). Compensatory strategies for reaching in stroke. Brain, 123 ( Pt 5) 940 953. Colebatch, J. G., & Gandevia, S. C. (1989). The distribution of muscular weakness in upper mo tor neuron lesions affecting the arm. Brain, 112 ( Pt 3) 749 763. Corti, M., McGuirk T.E., Wu, S.S., & Patten, C. (2010 ). Recovery of Upper Extremity Motor Function Post Stroke: Differential Effects of Two Therapeutic Interventions Manuscript accepte d for publication, Neurorehabilitation and Neural Repair. Cruz, E. G., Waldinger, H. C., & Kamper, D. G. (2005). Kinetic and kinematic workspaces of the index finger following stroke. Brain, 128 (Pt 5), 1112 1121. Davids, K., Button C., & Benett, S. (200 8). Dynamics of skill acquisition: A constraints led approach. Champaign, IL: Human Kinetics. Deffeyes, J. E., Harbourne, R. T., DeJong, S. L., Kyvelidou, A., Stuberg, W. A., & Stergiou, N. (2009). Use of information entropy measures of sitting postural sway to quantify developmental delay in infants. Journal of Neuroengineering and Rehabilitation, 6 Deffeyes, J. E., Harbourne, R. T., Kyvelidou, A., Stuberg, W. A., & Stergiou, N. (2009). Nonlinear analysis of sitting postural sway indicates development al delay in infants. Clinical Biomechanics, 24 (7), 564 570. Dewald, J.P., Pope, P.S., Given, J.D., Buchanan, T.S., & Rymer, W.Z. (1995). Abnormal muscle coactivation patterns during isometric torque generation at the elbow and shoulder in hemiparetic subj ects. Brain 118 (2), 495 510. Diedrich, F. J., & Warren, W. H., Jr. (1995). Why change gaits? Dynamics of the walk run transition. J Exp Psychol Hum Percept Perform, 21 (1), 183 202. Diedrichsen, J., Shadmehr, R., & Ivry, R. B. (2009). The coordination of movement: optimal feedback control and beyond. Trends Cogn Sci, 14 (1), 31 39. Dromerick, A. W., Edwards, D. F., & Hahn, M. (2000). Does the application of constraint induced movement therapy during acute rehabilitation reduce arm impairment after ischemi c stroke? Stroke, 31 (12), 2984 2988. Duncan PW, Stason WB, Adams HP, Adelman AM, Alexander DN, Bishop DS, Diller L,Donaldson NE, Grager CV, Holland AL. (1995). Post Stroke Rehabilitation: Clincal Practice Guideline. Rockville, Maryland. Duncan, P. W., Lai, S. M., & Keighley, J. (2000). Defining post stroke recovery: implications for design and interpretation of drug trials. Neuropharmacology, 39 (5), 835 841.

PAGE 125

125 Elbert, T., Ray, W. J., Kowalik, Z. J., Skinner, J. E., Graf, K. E., & Birbaumer, N. (1994). C haos and physiology: deterministic chaos in excitable cell assemblies. Physiol Rev, 74 (1), 1 47. Fleiss, J.L. (1986). Reliability of measurement. In. The design and analysis of clinical experiments (pp. 1 32). New York, NY: Wiley. Fuglmeyer, A. R., Ja asko, L., Leyman, I., Olsson, S., & Steglind, S. (1975). Post Stroke Hemiplegic Patient .1. Method for Evaluation of Physical Performance. Scandinavian Journal of Rehabilitation Medicine, 7 (1), 13 31. Glass, L. & Mackey, M.C. (1943). From clocks to chaos: The rhythms of life. UK: Princeton Goldberger, A. L., Bhargava, V., West, B. J., & Mandell, A. J. (1986). Some Observations on the Question Is Ventricular Fibrillation Chaos. Physica D, 19 (2), 282 289. Goldberger, A. L. (1997). Fractal variabilit y versus pathologic periodicity: complexity loss and stereotypy in disease. Perspect Biol Med, 40 (4), 543 561. Goldberger, A. L., Rigney, D. R., & West, B. J. (1990). Chaos and fractals in human physiology. Sci Am, 262 (2), 42 49. Gresham, G. E., Fitzpatr ick, T. E., Wolf, P. A., McNamara, P. M., Kannel, W. B., & Dawber, T. R. (1975). Residual disability in survivors of stroke -the Framingham study. N Engl J Med, 293 (19), 954 956. Grichting, B., Hediger, V., Kaluzny, P., & Wiesendanger, M. (2000). Impaired proactive and reactive grip force control in chronic hemiparetic patients. Clin Neurophysiol, 111 (9), 1661 1671. Haken, H., Kelso, J. A., & Bunz, H. (1985). A theoretical model of phase transitions in human hand movements. Biol Cybern, 51 (5), 347 356. H an, B., & Haley, W. E. (1999). Family caregiving for patients with stroke. Review and analysis. Stroke, 30 (7), 1478 1485. Harbourne, R. T., & Stergiou, N. (2009). Movement Variability and the Use of Nonlinear Tools: Principles to Guide Physical Therapist Practice Response. Physical Therapy, 89 (3), 284 285. Hermsdorfer, J., Hagl, E., Nowak, D. A., & Marquardt, C. (2003). Grip force control during object manipulation in cerebral stroke. Clin Neurophysiol, 114 (5), 915 929.

PAGE 126

126 Kamm, K., Thelen, E., & Jensen, J. L. (1990). A dynamical systems approach to motor development. Phys Ther, 70 (12), 763 775. Kaplan, D., Staffin, P. (1996). http://www.macalester.edu/~kaplan/hrv/doc/ Accessed December 21, 2009. Keele, S. W. (1968). Movement Control in Skilled Motor Performance. Psychological Bulletin, 70 (6p1), 387. Kelso, J. A. S. (1984). Phase Transitions and Critical Behavior in Human Bimanual Coordination. American Journal of Physiology, 246 (6), 1000 1004. Kurz, M. J., & Stergiou, N. (2003). The aging human neuromuscular system expresses less certainty for selecting joint kinematics during gait. Neuroscience Letters, 348 (3), 155 158. Lacquaniti, F., Soechting, J. F., & Terzuolo, S. A. (1986). Path constrain ts on point to point arm movements in three dimensional space. Neuroscience, 17 (2), 313 324. Lai, S. C., Mayer Kress, G., & Newell, K. M. (2006). Information entropy and the variability of space time movement error. J Mot Behav, 38 (6), 451 466. Lang, C. E., & Schieber, M. H. (2003). Differential impairment of individuated finger movements in humans after damage to the motor cortex or the corticospinal tract. J Neurophysiol, 90 (2), 1160 1170. Langhorne, P., Coupar, F., & Pollock, A. (2009). Motor recovery after stroke: a systematic review. Lancet Neurol, 8 (8), 741 754. Lashley, K.S. (1942). The problem of cerebral organization in vision. In Cattell, J. (Ed.), Biological Symposia Vol. VII. Visual mechanisms (pp. 301 322). Lancaster, PA: Jacques Cattell P ress. Latash, M. L., & Anson, J. G. (2006). Synergies in health and disease: relations to adaptive changes in motor coordination. Phys Ther, 86 (8), 1151 1160. Lexell, J. E., & Downham, D. Y. (2005). How to assess the reliability of measurements in rehabi litation. Am J Phys Med Rehabil, 84 (9), 719 723. Lipsitz, L. A., & Goldberger, A. L. (1992). Loss of 'complexity' and aging. Potential applications of fractals and chaos theory to senescence. JAMA, 267 (13), 1806 1809. Lodha, N., Naik, S. K., Coombes, S. A., & Cauraugh (2010) J. H. Force control and degree of motor impairments in chronic stroke. Clin Neurophysiol

PAGE 127

127 Mark, L. S., Nemeth, K., Gardner, D., Dainoff, M. J., Paasche, J., Duffy, M., et al. (1997). Postural dynamics and the preferred critical boun dary for visually guided reaching. J Exp Psychol Hum Percept Perform, 23 (5), 1365 1379. Matlab, R2009a, Natick, MA, USA. McIntosh, G. C., Brown, S. H., Rice, R. R., & Thaut, M. H. (1997). Rhythmic auditory motor facilitation of gait patterns in patients w ith Parkinson's disease. J Neurol Neurosurg Psychiatry, 62 (1), 22 26. Michaelsen, S. M., Jacobs, S., Roby Brami, A., & Levin, M. F. (2004). Compensation for distal impairments of grasping in adults with hemiparesis. Exp Brain Res, 157 (2), 162 173. Morass o, P. (1981). Spatial control of arm movements. Exp Brain Res, 42 (2), 223 227. Morris, D. M., Taub, E., & Mark, V. W. (2006). Constraint induced movement therapy: characterizing the intervention protocol. Eura Medicophys, 42 (3), 257 268. Nakayama, H., Jo rgensen, H. S., Raaschou, H. O., & Olsen, T. S. (1994). Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil, 75 (4), 394 398. Newell, K. M. (1976). More on Absolute Error, Etc. Journal of Motor Behavi or, 8 (2), 139 142. Newell, K.M. (1986). Constraints on the development of coordination. In Wade, M.G. & Whiting, H.T.A. (Eds.), Motor development in children: Aspects of coordination and control (pp. 341 360). Dordrecht, Netherlands: Martinus Nijhoff. N ewell, K.M & Slikfin, A.B. (1998). The nature of movement variability. In Piek, J.P. (Ed.), Motor behavior and human skill: A multidisciplinary perspective (pp.143 160). Champaign, IL: Human Kinetics. Newell, K.M., Deutsch, K.M., Sosnoff, J.J. & Mayer K ress, G. (2005). Variability in motor output as noise. In Davids, K, Bennet, S. & Newell, K. (Eds.), Movement System variability (pp. 3 23). Champaign, IL: Human Kinetics. Olsen, T. S. (1990). Arm and leg paresis as outcome predictors in stroke rehabilit ation. Stroke, 21 (2), 247 251. O' Sullivan SB, Schmitz TJ. (2000). Physical Rehabilitation Assessment and Treatment (4 th ed.). Davis Company; Philadelphia. 1168.

PAGE 128

128 Pandyan, A. D., Johnson, G. R., Price, C. I. M., Curless, R. H., Barnes, M. P., & Rodge rs, H. (1999). A review of the properties and limitations of the Ashworth and modified Ashworth Scales as measures of spasticity. Clinical Rehabilitation, 13 (5), 373 383. Pincus, S. M., Gladstone, I. M., & Ehrenkranz, R. A. (1991). A Regularity Statistic for Medical Data Analysis. Journal of Clinical Monitoring, 7 (4), 335 345. Pool, R. (1989). Is it healthy to be chaotic? Science, 243 (4891), 604 607. Portney, L.G. & Watkins, M.P. (2000). Foundations of clinical research: Applications to practice (2 nd ed .). New Jersey: Prentice Hall. Raghavan, P., Santello, M., Gordon, A. M., & Krakauer, J. W. (2010). Compensatory motor control after stroke: an alternative joint strategy for object dependent shaping of hand posture. J Neurophysiol, 103 (6), 3034 3043. Ra pp, P. E., Albano, A. M., Schmah, T. I., & Farwell, L. A. (1993). Filtered noise can mimic low dimensional chaotic attractors. Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics, 47 (4), 2289 2297. Reisman, D. S., & Scholz, J. P. (2006). Workspac e location influences joint coordination during reaching in post stroke hemiparesis. Experimental Brain Research, 170 (2), 265 276. Rossignol, S., & Jones, G. M. (1976). Audio spinal influence in man studied by the H reflex and its possible role on rhythmi c movements synchronized to sound. Electroencephalogr Clin Neurophysiol, 41 (1), 83 92. Sahrmann, S. A., & Norton, B. J. (1977). The relationship of voluntary movement to spasticity in the upper motor neuron syndrome. Ann Neurol, 2 (6), 460 465. Schmidt, R A. (1975). Schema Theory of Discrete Motor Skill Learning. Psychological Review, 82 (4), 225 260. Schmidt, R. A. (2003). Motor schema theory after 27 years: reflections and implications for a new theory. Res Q Exerc Sport, 74 (4), 366 375. Schmidt, R.A. & Lee, T.D. (2004). Motor control and learning: A behavioral emphasis.(4th ed.). Champaign, IL: Human Kinetics. Scholz, J. P. (1990). Dynamic pattern theory -some implications for therapeutics. Phys Ther, 70 (12), 827 843.

PAGE 129

129 Sethi, A., Patterson, T., There sa, M., & Richards, L.G. (2009). Movement in stroke more or less variability and its clinical relevance Society for Neuroscience, Chicago, IL. Shadmehr, R., & Mussa Ivaldi, F. A. (1994). Adaptive representation of dynamics during learning of a motor tas k. J Neurosci, 14 (5 Pt 2), 3208 3224. Shannon, C.E. & Weaver, W. (1949). The mathematical theory of communication. Urbana, IL: University of Illinois Press. Shea, C. H., Lai, Q., Wright, D. L., Immink, M., & Black, C. (2001). Consistent and variable pra ctice conditions: effects on relative and absolute timing. J Mot Behav, 33 (2), 139 152. Shelton, F. N., & Reding, M. J. (2001). Effect of lesion location on upper limb motor recovery after stroke. Stroke, 32 (1), 107 112 SIMM 4.2, MusculoGraphics, Inc., Santa Rosa, CA, USA. Slifkin, A. B., & Newell, K. M. (1999). Noise, information transmission, and force variability. J Exp Psychol Hum Percept Perform, 25 (3), 837 851. Smits Engelsman, B. C., Swinnen, S. P., & Duysens, J. (2006). The advantage of cyclic over discrete movements remains evident following changes in load and amplitude. Neurosci Lett, 396 (1), 28 32. Sosnoff, J. J., & Newell, K. M. (2006a). Aging, visual intermittency, and variability in isometric force output. Journals of Gerontology Series B Psychological Sciences and Social Sciences, 61 (2), P117 P124. Sosnoff, J. J., & Newell, K. M. (2006b). Are age related increases in force variability due to decrements in strength? Exp Brain Res, 174 (1), 86 94. Stergiou, N., Harbourne, R., & Cavanaugh J. (2006). Optimal movement variability: a new theoretical perspective for neurologic physical therapy. J Neurol Phys Ther, 30 (3), 120 129. Stergiou, N., Buzzi, U.G., Kurz, M.S. & Heidel, J. (2004). Nonlinear tools in human movement. In Stergiou (Ed.), Innovative analysis of human movement. (pp. 63 90). Champaign, IL: Human Kinetics. Sterr, A., Elbert, T., Berthold, I., Kolbel, S., Rockstroh, B., & Taub, E. (2002). Longer versus shorter daily constraint induced movement therapy of chronic hemiparesis: an exploratory study. Arch Phys Med Rehabil, 83 (10), 1374 1377.

PAGE 130

130 Tallis, R. (2003). The hand: a philosophical inquiry in human being : Edinburg, UK: Edinburg University Press. Taub, E., Miller, N. E., Novack, T. A., Cook, E. W., 3rd, Fleming, W. C., Nepom uceno, C. S., et al. (1993). Technique to improve chronic motor deficit after stroke. Arch Phys Med Rehabil, 74 (4), 347 354. Thach, W. T., Goodkin, H. P., & Keating, J. G. (1992). The cerebellum and the adaptive coordination of movement. Annu Rev Neurosci 15 403 442. Thaut, M.H., McIntosh, G.C., Prassas, S.G., & Rice, R.R. (1993). The effect of auditory rhythmic cuing on stride and EMG patterns in hemiparetic gait of stroke patients. Journal of Neurologic Rehabilitation 7, 9 16. Thaut, M, Kenyon, G. P., Schauer, M.L., & McIntosh, G.C. (1996) Rhythmic facilitation of movement sequencing: Effects on spatiotemporal control and sensory modality dependency. Music medicine. Edited by Pratt R, Spintge R. St. Louis, MMB pp 104 112. Thaut, M. H., Kenyon, G. P., Hurt, C. P., McIntosh, G. C., & Hoemberg, V. (2002). Kinematic optimization of spatiotemporal patterns in paretic arm training with stroke patients. Neuropsychologia, 40 (7), 1073 1081. Thaut, M. H., Kenyon, G. P., Schauer, M. L., & McIntosh, G. C. (1999). The connection between rhythmicity and brain function. IEEE Eng Med Biol Mag, 18 (2), 101 108. Thaut, M. H., McIntosh, G. C., & Rice, R. R. (1997). Rhythmic facilitation of gait training in hemiparetic stroke rehabilitation. J Neurol Sci, 151 (2), 2 07 212. Thaut, M. H., McIntosh, G. C., Rice, R. R., Miller, R. A., Rathbun, J., & Brault, J. M. (1996). Rhythmic auditory stimulation in gait training for Parkinson's disease patients. Mov Disord, 11 (2), 193 200. Theiler, J., Eubank, S., Longtin, A., Gal drikian, B., Farmer, J.D.,1992. Testing for nonlinearity in time series: the method of surrogate data. Physica D 58 (1 4), 77 94. Thirumala, P., Hier, D. B., & Patel, P. (2002). Motor recovery after stroke: lessons from functional brain imaging. Neurol R es, 24 (5), 453 458. Todorov, E., Shadmehr, R., & Bizzi, E. (1997). Augmented Feedback Presented in a Virtual Environment Accelerates Learning of a Difficult Motor Task. J Mot Behav, 29 (2), 147 158.

PAGE 131

131 Vaillancourt, D. E., & Newell, K. M. (2002). Changing co mplexity in human behavior and physiology through aging and disease. Neurobiol Aging, 23 (1), 1 11. van der Lee, J. H., Wagenaar, R. C., Lankhorst, G. J., Vogelaar, T. W., Deville, W. L., & Bouter, L. M. (1999). Forced use of the upper extremity in chronic stroke patients: results from a single blind randomized clinical trial. Stroke, 30 (11), 2369 2375. van Emmerik, R. E., Sprague, R. L., & Newell, K. M. (1993). Quantification of postural sway patterns in tardive dyskinesia. Mov Disord, 8 (3), 305 314. Van Thiel, E., Meulenbroek, R. G., Smeets, J. B., & Hulstijn, W. (2002). Fast adjustments of ongoing movements in hemiparetic cerebral palsy. Neuropsychologia, 40 (1), 16 27. Vanemmerik, R. E. A., Sprague, R. L., & Newell, K. M. (1993). Quantification of Post ural Sway Patterns in Tardive Dyskinesia. Movement Disorders, 8 (3), 305 314. Vereijken, B., Vanemmerik, R. E. A., Whiting, H. T. A., & Newell, K. M. (1992). Free(Z)Ing Degrees of Freedom in Skill Acquisition. Journal of Motor Behavior, 24 (1), 133 142. Vi con 612, Oxford Metrics, Oxford, UK. Winstein, C., Wing, M, & Whitall, J. (2003). Motor control and learning principles for rehabilitation of upper limb movements after brain injury. In Grafman, J. & Robertson, I.H. (Eds.), Handbook of Neuropsychology ( 2 nd ed.) (pp. 77 137). Amsterdam, Netherlands: Elsevier Science. Wolf, S. L., Winstein, C. J., Miller, J. P., Taub, E., Uswatte, G., Morris, D., et al. (2006). Effect of constraint induced movement therapy on upper extremity function 3 to 9 months after s troke The EXCITE randomized clinical trial. Jama Journal of the American Medical Association, 296(17), 2095 2104. Wolpert, D. M., Ghahramani, Z., & Jordan, M. I. (1995). Are arm trajectories planned in kinematic or dynamic coordinates? An adaptation stu dy. Exp Brain Res, 103(3), 460 470. Woodbury, M. L., Howland, D. R., McGuirk, T. E., Davis, S. B., Senesac, C. R., Kautz, S., et al. (2009). Effects of Trunk Restraint Combined With Intensive Task Practice on Post stroke Upper Extremity Reach and Function : A Pilot Study. Neurorehabilitation and Neural Repair, 23(1), 78 91.

PAGE 132

132 Wu, C. Y., Lin, K. C., Chen, H. C., Chen, I. H., & Hong, W. H. (2007). Effects of modified constraint induced movement therapy on movement kinematics and daily function in patients with stroke: a kinematic study of motor control mechanisms. Neurorehabil Ne ural Repair, 21(5), 460 466.

PAGE 133

133 BIOGRAPHICAL SKETCH Amit Sethi did bachelor s degree in occupational therapy from Manipal College of Allied Health Sciences, Karnataka, India He then worked for about two years imparting occupational therapy services for c hildren with neurological disabilities. He then moved he worked as an occupational therapist in the brain injury and stroke team at The Institute of Rehabilitation and Research at Houston, T exas His interest in knowing the intricacies of understanding the movement dysfunction following brain stroke encouraged him to pursue the Rehabilitation Sciences Doctor al program at University of Florida (UF). He was funded by the Alumni fellowship for four years of his graduate education and from the teaching assistant offered by the Department of Occupational Therapy at UF. His research employed biomechanical measures to understand the variability of upper extremity movement post stroke and was mentored under the expert tutelage of Dr. Lorie Richards. His future plans involve continuing research in the area of rehabilitation sciences to discover more potent therapies fo r individuals with neurological impairments. In the long term, Amit plans to become a successful educator as well as scientist and aims to establish a career in rehabilitation sciences based upon strong clinical inquiry and rationale.