HUMAN ACTIVITY RECOGNITION FOR ALZHEIMER’S PATIENTS USING MACHINE LEARNING

This project presents a hybrid plan recognition model, based on probabilistic description logic, which addresses the issues of recognizing the activities and the errors of Alzheimer's patients at an early stage of the disease. This model has been implemented with an embedded smart watch pair, (one on the wrist and other on the thighs for high accuracy) for patients with Alzheimer's disease which in turn offers assistance in task completion and to remind about the pending essential tasks. The patient have to wear a pair of bands(wrist & leg), from these bands accelerometer and gyroscope values are extracted and the activities like brushing, bathing, eating, going to toilet, taking medicine etc. can be tracked and reminded during right intervals. This result can be analysed through a dedicated mobile application which is under control by the bystander. The application can also remind the patients about those pending tasks which he/she haven't completed and are pending .The pending tasks are analyzed and are alerted to the bystander and also an alert is given to the patient to remind about the activity that he has forgot to do. This in turn would help to keep the activities in order and sometimes it may help to bring their memory back. The project focuses on providing complete assistance to Alzheimer's patients in their daily life.