UC San Diego

The landscape of health technologies is rapidly evolving and commercial wearabledevices equipped with health sensors offer a promising avenue for continuous health monitoring. The ubiquity of these devices among consumers presents an unprecedented opportunity to leverage the wealth of data that wearables collect for health applications. Despite their widespread use, there have only recently been developments towards utilizing these data for enhancing health monitoring. However, research stemming from recent efforts to gather large-scale longitudinal wearable datasets suggests these devices might hold potential in detecting the presence of and characterizing aspects of acute physiological changes. The application of wearable device data might be particularly useful in the context of acute physiological changes given wearables’ ability to monitor a number of physiological vital signs both passively and longitudinally. Passive monitoring uniquely enables longitudinal comparisons of individuals to themselves over time and real-time identification of significant deviations indicative of changes in health status. This thesis explores the application of data from wearable devices for detecting and characterizing physiological changes following significant health events, specifically vaccination for COVID-19 and the onset of fever. I also present the first comprehensive analysis of multiple large-scale, longitudinal wearable device datasets, therein assessing the generalizability of algorithms for monitoring acute illnesses and characterizing the biases in these datasets, some of which are correlated with demographic variables. Through this research, I demonstrate the potential of commercial wearable devices in enhancing our understanding and monitoring of acute physiological changes, and present a framework through which industry and research standards might emerge to speed the evolution of this field.