Robot-measured performance metrics in stroke rehabilitation

One of the useful features of robotic rehabilitation is the possibility of movement quantification, which is currently lacking in conventional rehabilitation therapy. Movement performance measures calculated from this quantitative information serves various purposes - (a) a good supplement to clinical assessment measures, (b) can be more sensitive than many clinical measures which use ordinal scales for scoring, (c) can be used to track a patient's recovery over time. Our research group has developed a 5 degree-of-freedom wearable exoskeleton robot for upper-extremity rehabilitation (RUPERT); RUPERT provides movement kinematics information in the form of joint angles, and also provides the pressure inside the pneumatic muscle actuators that drive the robot. In this paper we describe some useful robot-measured performance metrics that can be calculated from the sensor information collected from RUPERT. Some of the important performance-metrics described in this paper are - (a) Amount-of-assistance, (b) Smoothness, and (c) Movement synergy. We present a new method for calculating smoothness, which is uses a very different approach from some of the currently available approaches for calculating smoothness.; we call this approach the dasiaspectral methodpsila, which looks at the frequency spectrum of the movement velocity signal to estimate movement smoothness. In addition, we also present a method to analyze the effect of target location and DOF on the performance metrics, and also a method to detect fatigue.