Robotic Rehabilitation Gaming Strategies and Low-Dimensional Analysis of Hand Trajectories
- Author(s): Roldan, Jay Ryan Urbanozo;
- Advisor(s): Milutinovic, Dejan;
- et al.
An effective rehabilitation plan is a combination of a carefully thought out intervention strategies and a practical, reliable and sensitive recovery assessment method. With robot-aided therapy, these two components can be integrated into one complete system. In this thesis, we present the development of a rehabilitation software system that includes seven rehabilitation games and introduced a novel recovery assessment and tracking method using multidimensional scaling (MDS) for the purpose of extending the applicability of a 7-DOF upper limb exoskeleton device into a complete rehabilitation system.
The rehabilitation software is composed of three components: control, data acquisition, and rehabilitation games. It is implemented into a 7-DOF upper limb exoskeleton system that acts as a haptic device providing force feedback with a virtual environment and enforces proper arm posture. We developed seven rehabilitation games for the purpose of challenging motor control coordination and promoting neuromuscular recovery. The development effort successfully produced a research platform for investigating effective rehabilitation strategies providing novel insights into the difference of unilateral and bilateral training effectiveness, understanding of rehabilitation game design and evaluation, and uncovered evidence on the superiority of robot-aided therapy compared to conventional training.
Current state of the art assessment method such as the Fugl-Meyer and Wolf function test lacks the ability to objectively characterize stroke-impacted motions. The proposed recovery assessment and tracking method is a quantitative approach of identifying hand trajectory dissimilarity using multidimensional scaling (MDS). Using high-rate motion capture system, hand trajectories of both healthy and stroke-impacted hemiparetic subjects were captured. An MDS map was generated based on the mutual difference of two trajectories using area as a dissimilarity variable. The map reveals both structural and individual dissimilarities that presents quantifiable difference and variability of individual subjects. From this, we can identify and track the progress of recovery based on the difference of trajectory point from the cluster of healthy.