DAMON: A Data Authenticity Monitoring System for Diabetes Management
Abstract: This project designs a data authenticity monitoring system for use in an Internet of Medical Things (IoMT) system assembled to treat Type 1 Diabetes (T1D). We describe the use of Signal Temporal Logic (STL) for synthesizing and monitoring a range of system properties relevant to T1D treatment, including constraints on glycemic variability and insulin delivery. We perform retrospective analysis of posterior probabilities of multiple meal hypotheses to detect suspicious meal events. Using a corpus of patient clinical study data, we provide experimental results demonstrating the detection of system events indicative of compromised data authenticity.