The increasing demand for ongoing, remote health monitoring for people with chronic illnesses and aging populations calls for a change from passive data collecting to intelligent, proactive systems [1], [2]. Real-time, on-device predictive analytics, reliable fall detection, and an integrated, closed-loop emergency response system that connects users to emergency medical assistance are frequently absent from current commercial wearables. We introduce an integrated wearable system with an ADXL345 accelerometer for fall detection and a MAX30102 sensor for heart rate and SpO2 monitoring, all based on an ESP32 microcontroller. The solution uses an on-device AI model that is lightweight and tuned with TensorFlow Lite to detect anomalies in physiological data in real time. The system reliably detects abnormalities in vital signs and shows great efficacy in differentiating falls from activities of daily living (ADLs). Importantly, it automatically retrieves the user’s GPS coordinates and uses the Google Maps API to find local medical institutions, achieving an end-to-end emergency alert latency of less than 5 seconds. Real-time, on-device predictive analytics, reliable fall detection, and an integrated, closed-loop emergency response system that connects users to emergency medical assistance are freq By bridging the crucial gap between health anomaly detection and practical emergency intervention, our work offers a low-latency, energy-efficient, privacy-preserving approach that improves patient safety and autonomy.