Home-based health monitoring of the elderly through gait recognition

Abstract

In Europe, in particular, growing numbers of elderly people need sustainable elderly care, which the young are not able to provide. As an alternative, elderly care can be provided through home-based, automatic, health-monitoring systems. Here we propose data-mining algorithms in a system for the automatic recognition of health problems, activities and falls through the analysis of gait. The gait of the elderly is captured using a motion-capture system and the resulting time series of position coordinates are analyzed with a data-mining approach in order to classify it into five health states: 1 normal, 2 with hemiplegia, 3 with Parkinson's disease, 4 with pain in the back and 5 with pain in the leg, or into five activities/falls: 1 accidental fall, 2 unconscious fall, 3 walking, 4 standing/sitting, 5 lying down/lying. We propose and analyze four data-mining approaches: 1 CML --Classical machine-learning approach with raw sensor data, 2 SCML --Classical machine-learning approach with semantic attributes, 3 MDTW --Multidimensional dynamic time-warping approach with raw sensor data and 4 SMDTW --Multidimensional dynamic time-warping approach with semantic attributes. According to the results of the experiments, SMDTW achieved the highest classification accuracy of the four proposed approaches, and transforming the raw data into the semantic attributes significantly improved the performance of the approaches.Since the observed health problems are related also to postural instability and danger of falling, their early detection helps to prevent elderly people from falling.