Real-Time Anomaly Detection on Smart Bandages with Ultra-Low Power System-Embedded Machine Learning

This study delves into the concept of implementing machine learning algorithms within Ultra-Low Power (ULP) Systems for use in an intelligent bandage, designed to monitor local variations in temperature, moisture, and gas emissions in real-time. The proposed system hardware consists of a ULP microcontroller and several ULP sensor nodes. An n-class Classification Machine Learning algorithm has been trained and then deployed on the microcontroller. This innovative strategy aims to perform all computation/decision-making on the microcontroller, thus reducing system-level complexity and power consumption.