Robotic assessment of hand function with the HapticKnob

This paper investigates the assessment of hand function after stroke using the HapticKnob, an end-effector based robotic device to train grasping and forearm pronation/supination. A method to extract meaningful parameters to evaluate hand function from kinematic data recorded by the robot during rehabilitation exercises is presented. Step-wise regression analysis has been performed in an attempt to reconstruct clinical assessment scores from the kinematic data collected during a 6-week rehabilitation therapy with the HapticKnob Good correlations between clinical and reconstructed scores (r=0.669 for Fugl-Meyer Assessment, r=0.689 for Motricity Index, r=0.599 for Motor Assessment Scale, and r=0.792 for Modified Ashworth Scale) illustrate the potential of these objective measures to quantitatively evaluate hand motor function. This offers new possibilities to monitor patients' progress and customize exercise challenge during rehabilitation therapy.