Toward Improving Robotic-Assisted Gait Training: Can Big Data Analysis Help Us?

Abstract

Over the past years, in order to care neurolodical diseases, beyond conventional physical treatments, robotics rehabilitation has been widely adopted for improving the patients’ therapies. Recent scientific works were mainly aimed at personalizing treatments, according to the patient’s clinical conditions. On the contrary, this scientific work aims to propose an alternative approach based on big data analytics coming from the sensors of robotic rehabilitation devices in order to improve the patient’s therapy in the perspective of a healthcare Cloud of Things scenario. We perform an exploratory analysis considering big data coming from sensors installed in Lokomat, i.e., one of the major robotic rehabilitation devices, in order to study the data model and the predictors that will allow clinical operators to forecast the best treatment personalizing the therapy. Data analysis proves that there are moderate correlations among features referring to stance and swing biofeedbacks of hip and knee. From the analytical point of view, these values may approximate the stance and swing phases of knee and hip. This can be compared with the normal gait pattern of healthy individuals, so as to point out those patients having a closer normal ambulation. Obtained results are comparable with the Lokomat therapy outcomes of patients with neurological injuries by means of pattern recognition techniques.