Deep Convolutional Neural Network Classifier for Travel Patterns Using Binary Sensors

The early detection of dementia is crucial in independent life style of elderly people. Main intention of this study is to propose device-free non-privacy invasive Deep Convolutional Neural Network classifier (DCNN) for Martino-Saltzman's (MS) travel patterns of elderly people living alone using open dataset collected by binary (passive infrared) sensors. Travel patterns are classified as direct, pacing, lapping, or random according to MS model. MS travel pattern is highly related with person's cognitive state, thus can be used to detect early stage of dementia. The dataset was collected by monitoring a cognitively normal elderly resident by wireless passive infrared sensors for 21 months. First, over 70000 travel episodes are extracted from the dataset and classified by MS travel pattern classifier algorithm for the ground truth. Later, 12000 episodes (3000 for each pattern) were randomly selected from the total episodes to compose training and testing dataset. Finally, DCNN performance was compared with three other classical machine-learning classifiers. The Random Forest and DCNN yielded the best classification accuracies of 94.48% and 97.84%, respectively. Thus, the proposed DCNN classifier can be used to infer dementia through travel pattern matching.