Prithaj's Thesis Defense

When: June 22nd at 10:30 AM

Where: Innovation Hall E425, UVM

Quantifying Chronic Stress: A Reproducible Method for Measuring Allostatic Load with Wearable Sensors

Chronic stress is a leading contributor to long-term morbidity and mortality, with college students representing a particularly vulnerable demographic. The construct that most directly maps to these downstream outcomes — allostatic load (AL), the cumulative biological wear-and-tear from sustained stress exposure — has historically been operationalized through invasive clinical biomarker panels, limiting its measurement to cross-sectional snapshots. In parallel, the wearable stress literature has converged on detecting acute autonomic arousal in controlled laboratory settings, capturing transient state rather than cumulative trait. To bridge this gap, we analyze longitudinal sleep and heart-rate data collected using a consumer-grade wearable device (Oura ring) from a large cohort of college students. We derive two complementary measures of AL: a composite z-score capturing the magnitude of dysregulation, and a hidden Markov model capturing the shape of dysregulation over time. The former mirrors the canonical MacArthur operationalization in form; the latter exploits the temporal resolution of wearable data to recover dynamical features of AL that cross-sectional clinical assessment cannot resolve. Because a wearable accesses only the autonomic and sleep-regulatory signals of the construct, the measures it yields constitute the autonomic facet of allostatic load rather than the full multi-system panel. This approach provides a scalable, non-invasive method for measuring a facet of a construct previously accessible only through clinical assessment, and establishes a reproducible pipeline for future wearable-based biomarker research.

xkcd