Implementation of an AI-Powered Wearable for Remote Patient Surveillance

This article focuses on the design and implementation of a wearable AI-driven gadget for remote patient monitoring. It helps overcome the common obstacles in the traditional healthcare system, where timely action, accessibility, and constant supervision are always a challenge. With the integration of the Internet of Things, embedded systems, and AI, this wearable facilitates round-the-clock health monitoring. The hardware implementation uses a Seeed Studio XIAO ESP32-S3 microcontroller with a MAX30102 sensor for heart rate and SpO₂, measurements, an OLED display for real-time feedback, a buzzer for notifications, and a LiPo battery for portability. The software implementation involves embedded systems, Node.js, and PostgreSQL with real-time communication using Socket.IO. An LSTM neural network is used for anomaly detection and predictive health classification. Through experimental evidence, real-time surveillance is successfully performed with a latency of below 2 seconds and a model accuracy of 97%, and an accuracy of overlooking critical events of 99%. It was observed, however, that there were challenges like sensor instability in movement and limited battery life. The research finds that AI-based wearable technologies can be used to deliver scalable healthcare, especially in resource-limited settings, but need additional optimization and clinical trials.