Published in IEEE International Conference on Consumer Electronics and under review in IEEE Access

We develop an approach for systematically designing continuous monitoring solutions for early symptom diagnosis. Effective early diagnosis requires collecting and correlating symptoms derived from a number of vitals. For designing a continuous monitoring solution, it is crucial to determine the vitals to be monitored for targeted detection, the errors that can be tolerated, various parameters that need to be tuned, etc. Furthermore, this determination must be made before the design of the monitoring solution itself.  Our approach shows how to use a variety of machine learning techniques to systematically derive, tune, and optimize the vitals to be monitored before accessing the continuous monitoring data.  We show the effectiveness of our approach in the design of a wearable for early detection of COVID-19 infections in symptomatic patients.