This proposed project aims to

• compare the effects of dose-matched unilateral vs bilateral vs unilateral combined with bilateral upper limb rehabilitation based on task-related practice
• compare the effects of unilateral vs bilateral training based on robot-assisted devices
• study the predictors of treatment outcomes and clinimetric properties of the biomechanical measures

An estimated 30% to 66 % of stroke victims have poor upper extremity function at 6 months after stroke, which contributes to long-term disability in these patients. Training the affected UE of stroke patients has thus been a mainstay of neurorehabilitation. The goal of the intervention is to restore motor and daily functions. New therapeutic strategies have been developed based on principles of neurorehabilitation and motor learning. The need for comparative effectiveness research of the innovative treatments has been called for to promote evidence-based practice and translational science in stroke motor rehabilitation. Wolf and Whitall called for rigorous comparisons between unilateral and bilateral training. McCombe Waller and Whitall also indicated the importance of combining unilateral with bilateral training. For low-functioning patients, robot-assisted training is a possible approach. Identifying possible determinants for unilateral and bilateral training outcomes may also elucidate the factors that influence treatment outcomes. Outcome measures are required that have good clinimetric properties for measuring the effects of the rehabilitation strategies.

The long-term objective is to conduct comparative efficacy research to identify the possible mechanisms and the relative effectiveness of existing rehabilitation approaches of fully defined-, evidence-based, and theory-grounded approaches. This proposed project aims to (1) compare the effects of dose-matched unilateral vs bilateral vs unilateral combined with bilateral upper limb rehabilitation based on task-related practice; (2) compare the effects of unilateral vs bilateral training based on robot-assisted devices; and (3) study the predictors of treatment outcomes and clinimetric properties of the biomechanical measures. The outcome measures will span the spectrum of health-related functioning, including motor and neural control, movement performance, daily functions, and quality of life. Motor and neural control mechanisms involving movement strategies, muscle and force output, and brain reorganization will be evaluated using kinematic, electromyographic (EMG), and kinetic analyses, and also functional magnetic resonance imaging (fMRI) examinations. Clinical outcomes will include movement performance, daily functions, and quality of life, measured by MYOTON-3, Modified Ashworth Scale, the Fugl-Meyer Assessment, Wolf Motor Function Test, Functional Independence Measure, ABILHAND questionnaire, accelerometers, and Stroke Impact Scale. This 5-year project will recruit an estimated 200 patients with stroke. For Part 1, 120 patients with mild-to-moderate motor impairment will be randomized to the dose-matched unilateral, bilateral, or combined group. For Part 1, 80 stroke patients with moderate motor impairment will be randomized to the dose-matched unilateral or bilateral robot-assisted therapy group. Treatment regimens will be designed to ensure that patients in the 3 groups in Part 1 and 2 groups in Part 2 receive an equivalent intensity of treatment (5 days/week for 1.5 hours/day for 4 consecutive weeks). The intervention will be provided at 4 hospitals. Two certified occupational therapists will be trained in the administration of the 5 types of rehabilitation protocols by the PI and a co-PI and will complete a written competency test before subject treatment.

Biomechanical (kinematic, EMG, and kinetic) and fMRI examinations will be performed before, immediately after the 4-week intervention period, and at the 3-month follow-up period (no fMRI examination at follow-up). Clinical measures will be administered before, at midterm (2 weeks after intervention), immediately after, and 3 months after the intervention. At least 3 examiners (1 research assistant, 1 graduate student, and 1 post-doc fellow) blind to group allocation will be in charge of biomechanical, fMRI, and clinical measures. Each type of measure will require 2 evaluators to collaborate for completing the assessment. Before being allowed to work with participants, the examiner's competence will be assessed by the PI and co-PI, and interrater reliability for clinical tests established.

Multivariate analysis of covariance will be used to examine changes in all outcome measures as a function of treatment while controlling for baseline data. Multiple regression models will be established to determine the potential predictors for different functional outcomes of each intervention (unilateral and bilateral training approach). The responsiveness and validity of the biomechanical measures relevant for clinimetric scrutiny will be examined by the index of standardized response mean and Spearman correlation coefficients. The clinical important differences of the biomechanical parameters will be determined by the anchor-based and the distribution-based clinical important differences estimate.

Inclusion Criteria:

• 40 to 75 years old
• 6 to 24 months after stroke onset from a first-ever unilateral stroke
• An initial 26 to 56 scores on the UE subsection of the
• Premorbid right-hand dominance evaluated by the Edinburgh Handedness Inventory
• No excessive spasticity in the shoulder and elbow joints of the affected UE (Modified Ashworth Scale score ≤ 3 in each joint)
• Sufficient cognitive ability, defined as a score of more than 24 on the Mini Mental State Examination

Exclusion Criteria:

• Physician-determined major medical problems or poor physical condition that would interfere with participation
• Excessive pain in any joint that might limit participation