Assessing REM Sleep as a Biomarker for Depression Using Consumer Wearables

Area of Science:

• Neuroscience
• Sleep Medicine
• Digital Health

Background:

• Rapid-eye-movement (REM) sleep disinhibition, characterized by shorter REM latency and a larger REM fraction, is a known laboratory marker for major depression.
• The feasibility of capturing these REM sleep patterns using consumer wearables in daily life is not well-established.

Purpose of the Study:

• To assess if REM latency and REM sleep coefficient, derived from Apple Watch data, can be efficiently measured in everyday settings.
• To examine the association between wearable-derived REM sleep metrics and depressive symptom severity.

Main Methods:

• 191 adults recorded sleep using an Apple Watch for 15 nights, with data streamed via a custom iOS app.
• Sleep stages were determined using a validated neural-network model; REM latency and REM sleep coefficient were calculated.
• Depressive severity was assessed using the Beck Depression Inventory (BDI).

Main Results:

• Mean BDI score was 13.52 ± 6.79, REM sleep coefficient was 24.05 ± 6.52%, and REM latency was 103.63 ± 15.44 minutes.
• REM latency showed a significant negative correlation with BDI scores (ρ = -0.673, p < 0.001).
• REM sleep coefficient demonstrated a significant positive correlation with BDI scores (ρ = 0.678, p < 0.001).

Conclusions:

• Wearable-derived REM latency and REM proportion collectively explain a substantial portion (62%) of the variance in depressive symptom severity.
• These REM sleep metrics show promise as accessible digital biomarkers for depression.
• Further longitudinal and interventional studies are warranted to explore therapeutic potential through modifying REM architecture.