Trait-like nocturnal sleep behavior identified by combining wearable, phone-use, and self-report data

Area of Science:

• Sleep Science
• Digital Health
• Behavioral Science

Background:

• Polysomnography (PSG) is accurate but difficult to scale for long-term sleep behavior characterization.
• Consumer sleep trackers and smartphone data offer potential for scalable, real-world sleep monitoring.

Purpose of the Study:

• To assess the feasibility and utility of integrating consumer sleep trackers, ecological momentary assessment (EMA), and phone interaction data for characterizing habitual sleep behavior.
• To determine if discrepancies in multi-modal sleep data could reveal stable interindividual differences.

Main Methods:

• Integrated sleep measurements from consumer sleep trackers, smartphone EMA, and user-phone interactions in 198 participants over 2 months.
• Analyzed agreement between modalities and used k-means clustering to identify patterns in discrepant data.

Main Results:

• High user retention (>80%) and strong agreement (rho = 0.81-0.92) in bed and wake times across modalities.
• Discrepancies exceeding 1 hour on ~23% of nights revealed three distinct, stable interindividual patterns.
• These patterns were systematically associated with age, sleep timing, time in bed, and peri-sleep phone usage.

Conclusions:

• Multi-modal consumer-based sleep tracking is feasible and scalable for characterizing population sleep behavior.
• Data discrepancies, rather than being problematic, are valuable for identifying stable individual differences in sleep and peri-sleep behaviors.
• This approach enhances the understanding of population sleep dynamics beyond traditional methods.