Physical Activity Monitoring for Assisted Living at Home

We propose a methodology to determine the occurrence of falls from among other common human movements. The source data is collected by wearable and mobile platforms based on three-axis accelerometers to measure subject kinematics. Our signal processing consists of preprocessing, pattern recognition and classification. One problem with data acquisition is the extensive variation in the morphology of acceleration signals of different patients and under various conditions. We explore several effective key features that can be used for classification of physical movements. Our objective is to enhance the accuracy of movement recognition. We employ classifiers based on neural networks and k-nearest neighbors. Our experimental results exhibit an average of 84% accuracy in movement tracking for four distinct activities over several test subjects.