Detecting Activities of Daily Living for the Elderly Using Magnet Sensors

In recent years, the elderly population has been growing significantly, leading to an increased need for efficient monitoring systems to ensure their well-being and safety. This paper addresses the issues of reliability, maintenance, and privacy found in previous studies on activity recognition systems for the elderly. We propose a prototype system based on magnetic sensors, running on a Raspberry Pi 3 and requiring 3v energy. This system resolves the identified issues by leveraging the unique properties of magnetic sensors, such as their insensitivity to external interferences, unlimited service life, and high level of reliability. Furthermore, the prototype stores only binary data, ensuring privacy protection. We conducted a comparative evaluation with existing prototypes and observed that our system successfully detected all expected events and activities with high reliability and minimal maintenance. Future work will involve integrating all modules, deploying the system in a real-life environment, and continuously evaluating its performance for further improvements.