Prediction of response to robot-aided motor neuro-rehabilitation of children with cerebral palsy

Robot-aided neuro-rehabilitation is increasingly being incorporated into rehabilitation practices. The aim of this study was to search for patterns in the data acquired by a robot in the baseline evaluation session which could predict progress over the next therapy sessions. Kinematic and kinetic data were acquired during robot-mediated evaluation sessions of 12 children with motor impairments due to hemiplegic cerebral palsy (CP). Time and wavelets features were extracted from the data and used for K-means clustering. The data were labeled by the Quality of Upper Extremity Skills Test (QUEST) and the gradient of improvement change in the QUEST between the baseline assessment and therapy follow-up one month after completion of 16 robot-mediated therapy sessions. Two distinct clusters segregated these 12 children into performers and non-performers in terms of the QUEST.