Prediction Scoring in Exergames for Rehabilitation Patients using K-Means Clustering

Abstract

This paper highlighted a prediction scoring of difficulty modes in Medical Interactive Recovery Assistant (MIRA) exergames based on Kinect-based Rehabilitation Gaming System (RGS) for rehabilitation patients. The case study uses 19 rehabilitation patients with different lower limb limitations caused by stroke, traumatic brain injury (TBI) and spinal cord injury (SCI). MIRA exergames consists of three difficulty modes which are easy, medium and hard. Currently, physiotherapist will decide on difficulty mode based on the patients’ improvement and most of the time they will used the default setting for every patient playing exergames. Thus, this study proposed a new prediction scoring using k-means clustering algorithm to help suggesting the difficulty mode of the game. K-means clustering also is used to find the benchmarks of the patients’ history. The performance of the K-Mean algorithm is to make sure the patients are comfortable with their weakness side as suggested.