During the 3-week BMT period, patients will play a simple computer game (e.g., picking up apples) that tracks patient movements to provide instant visual feedback. The therapist will also provide verbal feedback. In each training session, the patient will practice 600 to 800 repetitions of passive-passive mode for 15 minutes, 600 to 800 repetitions of active-passive mode for 15 to 20 minutes, and 150 to 200 repetitions of active-active mode for 5 minutes, respectively, for the forearm pronation-supination and the wrist flexion-extension.

During the 3-week BAT period, participants will receive training in tasks focusing on bilateral symmetric or alternating movements of both ULs, such as lifting two cups, picking up two pegs, grasping and releasing two towels, and wiping the table with both hands.

Participants in CR group will receive therapist-mediated functional tasks training based on neurodevelopmental techniques with emphasis on functional tasks and muscle strengthening that will involve (1) active/passive range of motions, abnormal muscle tone/reflex inhibition techniques (such as stretching and/or weight bearing), and strengthening of affected arm for 15 to 20 minutes, (2) functional tasks for training fine motor or manual dexterity for 20 minutes, and (3) activities of daily living or functional tasks involving gross motor activities, coordination tasks, transition, mobility, and posture/balance for 15 to 20 minutes. The activities will be individualized according to the level of impairment and the needs of the patients.

The 10MWT assess mobility function by measuring the time and the numbers of strides required to walk 10 meters under two conditions: (1) with the self-pace of each participant (self-pace); (2) with the speed that the participants walked as soon as possible. The velocity and stride length of the participant are calculated. Research has validated the 10MWT in measuring mobility in stroke.

The activity monitors will be used to provide a direct and objective measure of the amount of the impaired arm movement outside the laboratory. This measure, quantitatively recording the amount of activity in free-living conditions, will be used to reflect increase in the amount of affected arm use over time. The participants will be asked to wear an activity monitor, Actigraphy (Ambulatory Monitoring Inc., New York), on each wrist for 3 consecutive days to measure what amount they actually do in their daily life before and immediately after treatment. The Actigraphy can be attached to the subject's limb and measures the motion of that limb through an accelerometer. In this project, acceleration will be sampled at 10 Hz and summed over a user-specified epoch. The recording epoch will be 2 seconds; recording capacity will be approximately 72 hours. The ratio of affected to non-affected arm movement is a reliable and valid real-life measure of treatment outcome.

The FIM consists of 18 items grouped into 6 subscales measuring self-care, sphincter control, transfer, locomotion, communication, and social cognition ability. Each item is rated from 1 to 7 (maximum score 126) based on the required level of assistance to perform the tasks (e.g., 1-complete assistance and 7-complete independence). A higher score on any subscale indicates a less disability. The FIM has been shown to have good inter-rater reliability, construct validity, and discriminant validity.

The MAL consists of 30 structured questions to interview how the patients rate the frequency (amount of use subscale) and quality (quality of movement subscale) of movements while using their affected arm to accomplish each of the 30 daily activities. The score of each item ranges from 0 to 5, and the higher scores indicate more frequently used or higher quality of movements. The summary score is the mean of the item scores. The clinimetric properties of the MAL in stroke patients have been validated.

The ABILHAND Questionnaire will be used to evaluate the ability of the UL in functional activities. It consists of 23 bimanual activities that measure subjectively perceived difficulty in performing some common activities in daily living, such as buttoning, cutting nails, and opening a bottle. The scale ranges from 0 to 3 (0 = cannot perform, 1 = performs partially or with great difficulty, 2 = performs with some difficulty, 3 = performs fully). Its reliability and construct validity has been confirmed.

he NEADL consists of 22 items scored on the basis of the requirement for help in performing the activity. There are four subscales (mobility, domestic, leisure and kitchen) and a total score. The scores range from 0 to 22, with higher scores representing better function. The psychometric properties of the NEADL have been validated in stroke patients.

The SIS 3.0 is a stroke-specific instrument of health-related quality of life. It contains 59 items measuring 8 domains (i.e., strength, hand function, activities of daily living/instrumental activities of daily living, mobility, communication, emotion. memory and thinking and participation) with a single item assessing perceived overall recovery from stroke. Items are rated on a 5-point Likert scale with lower scores indicating greater difficulty in task completion during the past week. The reliability, validity, and responsiveness have been shown to be satisfactory in stroke patients.

Proinflammatory markers and anti-inflammatory markers will be used to monitor changes in inflammation. Proinflammatory markers include C-reactive protein (CRP), tumor necrosis factor-α (TNF-α), interleukin (IL)-1α, and IL-6. Anti-inflammatory markers include IL-10 and transforming growth factor-β1 (TGF-β1). Oxidative stress and antioxidative markers will be used to monitor the changes on oxidative stress. Oxidative stress makers include 8-hydroxy-2-deoxyguanosine (8-OHdG) and F2-isoprostances. The antioxidants will include antioxidants superoxide dismutase (SOD) and glutathione peroxidase (GPx). We have no directional hypotheses regarding the changes of biomarkers because of its early stage in stroke rehabilitation research.

The kinematic analysis will involve unilateral and bilateral tasks, in which the participants will be asked to perform by using the affected upper limb or both upper limb simultaneously. A motion analysis system with 7 cameras (VICON MX 30d, Oxford Metrics Inc., Oxford, UK) will be used to capture the motion of arm (s) in kinematic testing. The markers will be attached on the styloid processes of the ulna. Depending on the unilateral or bilateral tasks, the makers will be placed on the affected arm or the both arms, respectively.