Combining biosignals to assess and monitor VR-assisted rehabilitation of children with Cerebral Palsy: a machine learning approach*

Cerebral palsy (CP) is one of the most frequent causes of motor disability in children. Physical therapy, coupled with sensing technology, assesses the integrity of the nervous system and muscular activity and eventually adapts the therapy to the patient’s abilities. In the TELOS Project, we carried out a single-blind controlled trial involving children with CP aged 6 to 16 years for VR-assisted rehabilitation of upper limbs and stability with tailored Serious Game.In collaboration with healthcare professionals, we used validated clinical scales and wearable sensors for surface electromyography (sEMG), electrocardiography (ECG), and motion capture to collect data from children before, and after therapeutic sessions. We aimed to integrate the acquired data to obtain a series of predictors to describe the therapy progression. We modelled the data using linear regressions with LASSO regularization to extract the relevant features. Although preliminary, our findings indicate heart activity as crucial information, suggesting the importance of emotions in the success of rehabilitation.Clinical Relevance: Gathering physiological data could allow therapists to perform tailored and adapted rehabilitation to the real skills and performance of the children.