Smart-Care: Innovative Multi-Sensor Technology for Elderly and Disabled Health Management

This project introduces an innovative e-health framework for monitoring health and daily activities through real-time data from scalar sensors and cameras. The framework focuses on Human Activity Recognition (HAR) using inertial data and activity detection via multimedia analysis. It features a two-stage heterogeneous architecture that integrates multimedia and scalar sensor data for enhanced anomaly detection, along with machine learning and deep learning models for high-precision activity recognition. Additionally, the Multivariate Anomaly Detector (MaD) identifies abnormal behavior by analyzing combined sensor readings, such as activity and heart rate. Data will be stored on a local server for extended periods to uncover subtle patterns and trends, with predictive analytics aiding in early anomaly detection. A prototype system, or testbed, will integrate these techniques, and the framework's efficiency will be validated using both existing and simulated datasets.