Senior Thesis Project

Consumer sleep wearables promise to help people sleep better. But there's a problem: we don't actually know if the metrics these devices track matter for real-world sleep quality.

My thesis investigates this gap using data from the Adolescent Brain Cognitive Development (ABCD) Study, tracking 179 adolescents who wore Fitbit Charge 2 devices alongside caregiver-reported sleep assessments.

I tested whether 33 objective sleep parameters—efficiency, duration, architecture, circadian timing, heart rate variability—could predict subjective sleep quality ratings. The results were striking: almost no correlation.

This fundamental disconnect suggests that wearables capture something real about physiological sleep processes, but miss what actually matters for how people experience their sleep. The device says you slept fine; the human says otherwise.

The implications matter for how we think about sleep technology, health monitoring, and the gap between what algorithms optimize and what humans actually care about.