Intention-driven robotic hand rehabilitation system with individuated finger training feature

Abstract

Stroke is one of the most prominent causes of disability in the world and is relatively prevalent as well. Many individuals become hemiplegic as a result of stroke; and in many cases, they become very dependent to others and require long term care and rehabilitation. Loss of hand function as a result of stroke is one of the most impactful consequences that hinder stroke survivors from doing their activities of daily living (ADL) by themselves. Different rehabilitation techniques have been developed to tackle this issue; while some studies have shown consistent results to improve arm function, not many have consistently shown hand and finger functions recovery (Langhorne et al. 2009). Constraint-induced movement therapy (CIMT) and robot-assisted training are the only two so far that are considered more promising (Langhorne et al. 2011). CIMT, however, has been criticized for only able to cater very selective population of stroke survivors with higher level of residual function. Robot-assisted rehabilitation, on the other hand, has focused more on larger and more proximal joints, such as shoulder and elbow, due the technical difficulties to facilitate more degrees of freedom (DOFs) required in the more distal joints like the fingers. This study aimed to: (1) extend the hand exoskeleton robot system previously developed by our group to allow force assessment and control of individual finger, (2) investigate finger characteristics after stroke, and (3) conduct a pilot randomized-controlled trial (RCT) to investigate the efficacy of the training using the system. For those purposes, here we propose the development of a hand exoskeleton robot with individual finger feature to cater the need of having a robot-assisted rehabilitation with hand gestures training. The device is a 5-DOF hand exoskeleton robot with 5 independent linear actuators, capable of facilitating individual finger movement assistance whenever necessary. The device is equipped with force sensors to measure MCP and PIP joint moments of each finger. The stability, linearity, and reliability of the joint moments measurement was tested and verified. A preliminary study with six right hemiplegic, right-hand dominant, stroke survivors and six age-matched neurologically intact control subjects was conducted to provide brief information about muscle weakness and finger individuality after stroke. Results showed decreased muscle strength and finger individuality post stroke. Additionally, it was also revealed that finger flexion individuality correlated very well with the functional ability reflected by the clinical scores (Pearson's r>0.9 for its correlations with ARAT and WMFT scores). In the next part of my study, a pilot randomized-controlled trial with 19 chronic stroke subjects (14 males and 5 females, aged 53.2 ± 9.9 years old) was conducted. The subjects were randomly distributed into two groups: the robot-assisted (robot) group or the non-assisted (control) group. Each subject was required to complete 20 one-hour sessions of the designated training. All subjects, regardless of the grouping, were to do the same task: moving a sponge with three different grips, i.e. hand grasp, three-finger pinch, and two-finger pinch. The robot group received intention-driven assistance from the device, while the control group was given no assistance as the linear actuators were detached from the device. The results showed that the robot group maintained its significant improvement of hand and upper limb functions 6 months after the training, indicated by improved ARAT (mean change = 14.00±5.75, p=0.044) and FMA-SE (mean change = 3.44±2.01, p=0.020) scores, while the control group did not show any significant improvement at the same time point. This suggested the potential efficacy of the training and the feasibility of using it in a clinical setting for after stroke upper limb rehabilitation. In the future, we are also exploring the possibility of using this training as a complement to CIMT, apart from being a standalone upper limb rehabilitation technique; allowing individuals with stroke to receive this intervention first when their functional ability has not match the requirement of CIMT and possibly enroll in CIMT provided that they improved enough following the proposed robot-assisted fingers rehabilitation.