A predictive positioning system using supervised learning for home care of older people

Position monitoring of older people inside a home is important to deliver appropriate services to meet their needs and preferences. In this paper, we compared five supervised learning algorithms that were used to monitor the position of older people whether inside or outside the room. A number of 1038 data records was obtained through 30 days behaviour monitoring of older people living independently at home. We compared Decision Tree, Support Vector Classification (SVC), Probabilistic Neural Network (PNN), Multilayer Perceptron (MLP), and Deep Learning algorithms. Deep Learning showed the best result with mean of validation accuracy equals to 72.8%.