No abstract

Manual observation in measuring and assessing stroke patient progress in fine motor rehabilitation will lead to inconsistencies especially when the patient is evaluated by different therapists or attends different rehabilitation facilities. In addition, it also increases therapist workload if they need to supervise many patients at the same time. Thus, a model was proposed to capture finger data from motion sensor device using Time-Based Simplified Denavit-Heartenberg (TS-DH) and the Finger State progress (FSP) model. Actual finger movement was compared with patterns of finger state for real-time evaluation of finger movement progress. The model will assist therapists in real-time or post-exercise evaluation of patient progress and analysis can be done during stroke rehabilitation exercise. As a conclusion, the model can be used efficiently in virtual stroke rehabilitation as real-time indicator or as a long term analysis to compare prior progress.

Virtual finger model is commonly used in many applications for stroke fine motor rehabilitation especially in Virtual Reality (VR) applications. Capturing movement data for fingers is one of the important phases in any virtual fine motor rehabilitation process. Manual observation provides inconsistent evaluation given by different therapists for different rehabilitation sessions. Although the process of capturing data is performed, without time series of captured data, the data will not have a significant impact in producing consistent and meaningful evaluation. Furthermore, the consistency of the assessment of rehabilitation sessions will deliver more robust rehabilitation progress analysis. This data is very useful when paired with time information which can be analyzed to produce optimal evaluation. This paper proposes Time-based Simplified Denavit-Heartenberg Translation (TS-DH) consisting of forward kinematic with simplified DH parameter for capturing coordinate of end of each bone from virtual finger model paired with timeframe data. The DH model is enhanced by implementing 2 additional rules in assigning joint parameter. The data will be recorded with timeframe of every finger movement. As a conclusion, TS-DH model can be used in any virtual finger environment accurately.