This research aims to integrate and develop a novel Brain-Computer Interface (BCI) controlled soft robotic glove, evaluate the ability of the glove in achieving common hand grasping postures and to assess the efficacy of the glove in assisting stroke patients with completing functional tasks. The BCI-controlled soft robotic glove utilizes patients' user intent to deliver specific electroencephalographic patterns that can trigger robot-assisted hand movement to the affected hand.

Hand motor impairment is very common after a stroke. These impairments include difficulty moving and coordinating the hands and fingers, which inhibits stroke patients from being able to perform daily functional tasks independently, resulting in a reduced quality of life. More than half of people with upper limb impairment after stroke will still have problems many months to years after their stroke. Therefore, improving hand function is a core element of rehabilitation. Many possible interventions have been developed; these may involve different exercises or training, specialist equipment or techniques, or they could take the form of a drug (pill or injection) given to help hand movement. There is limited evidence that suggests the following interventions may be effective: constraint-induced movement therapy, mental practice, mirror therapy,interventions for sensory impairment, virtual reality and a relatively high dose of repetitive task practice. Current hand rehabilitation robotic devices are typically driven by rigid linkages or joints, which subject the patient's fingers into a single plane of motion that will feel unnatural and uncomfortable. On top of that,these devices belong to the class of continuous passive motion (CPM) devices that only promote hand range-of-motion, but do not require the patient to play a semi-active role in performing the hand exercises. Furthermore, there is a huge demand for solutions assisting stroke patients with using the densely paralyzed hand to perform activities of daily living (ADL) in real life, which is not available at present. Most of the hand rehabilitation robotic devices available in the market cannot assist paralyzed hand to carry out ADL. To develop an assistive device to solve this unmet need, we decided to combine BCI technology with the wearable soft robotic glove, which enables actuation of paralyzed hand by motor imagery.

Inclusion Criteria:

• Aged 55-90 years regardless of lesion size, race
• History of stroke less than 3 months prior to participation
• Stroke type: ischemic or haemorrhagic
• Fugl-Meyer motor score (FM score) of upper extremity impairment of 0-45 out of a maximum score of 66 on the Fugl-Meyer assessment scale
• Ability to pay attention and maintain supported sitting for 1.5 hours continuously
• Lack of or poor hand mobility (Medical Research Council Grade ≤ 2/5)
• Able to give own consent
• Able to comprehend and follow commands (Abbreviated Mental Test equal or more than 7)
• Fulfil BCI resting brain states on initial screening.
• Unilateral upper limb impairment

Exclusion Criteria:

• Recurrent clinical stroke
• Functional status: severe aphasia or inattention, unstable medical conditions which may affect participation (e.g. unresolved sepsis, postural hypotension) or anticipated life expectancy of <1 year due to malignancy or neurodegenerative disorder)
• Hemispatial neglect (visual or sensory) or severe visual impairment despite visual aids
• History of severe depression or active psychiatric disorder
• Skull defect or previous cranial surgery as this would affect physical fit of EEG cap interface
• Local arm factors: severe spasticity Modified Ashworth scale >1+ in any region, fixed joint contractures or joint replacements, patients with poor skin conditions which would contraindicate repetitive arm training, upper limb pain impeding movements with visual analogue scale (VAS score) >4/10, other conditions ensuing upper limb weakness, skull defect, polydactyly or amputation of fingers, and allergy to electrodes or adhesive gel

The blinding will be carried out in such a way that the therapist will provide conventional therapy and assess the arm function without knowing the group that the patients belong to. The BCI Robot and Passive Robot will be administered by other study team member. A third party study team member not participating in the intervention will keep the log book which contains information regarding patients study ID and group. Randomization codes will be broken when all interventional and follow-up sessions were completed or when the patient decides to quit participation. The third party team member will unmask according to PI's request in this occasion.