IoT-Enabled Wearable for Continuous Elderly Health Monitoring and AI-Based Emergency Detection

This paper is an IoT-enabled standalone wearable that integrates with continuous real-time biosignal processing and AI-driven scream detection that involves emergency identification in elderly individuals. This system avails heart rate (MAX30102), and vibration sensors interfaced with Raspberry Pi Pico 5. We use the Mel-Frequency Cepstral Coefficients (MFCCs) and CNN-SVM classifiers, which identifies distress vocalizations and predicts emergencies. Wi-Fi transmission ensures celerity alert delivery, while tactile sensors adds as a manual trigger adds redundancy. This paper presents hardware-software co-design, AI integration, Shannon's theorem-based encoding, and Nyquist rate optimization, demonstrating its efficacy in real-world elderly care applications. This system with (MFCCs) and CNN-SVM model which processes the model with an accuracy between 94% to 97%.