Exploring Variations in Sleep Perception: Comparative Study of Chatbot Sleep Logs and Fitbit Sleep Data

Area of Science:

• Digital Health
• Sleep Science
• Human-Computer Interaction

Background:

• Patient-generated health data (PGHD) from wearable devices are increasingly vital for disease management.
• Wearables like Fitbits can track sleep metrics (e.g., sleep onset, offset, total sleep time), but have limitations in accuracy and depth of sleep assessment.
• Subjective patient reporting is essential and cannot be fully replaced by wearable device data.

Purpose of the Study:

• To analyze the characteristics of individual time perception in sleep compared to wearable device data.
• To understand how users perceive and report sleep times versus objective measurements.
• To inform the effective use of time-related PGHD, specifically sleep data.

Main Methods:

• Collected sleep data over two weeks using Fitbit devices.
• Gathered daily sleep records from participants via chatbot conversations.
• Statistically compared Fitbit-measured sleep data with self-reported data from chatbot interactions.

Main Results:

• Analysis included 543 participants (aged 30-59) with over 7 days of data.
• Participants often reported sleep times in hourly or 30-minute increments, generally within 60-90 minutes of Fitbit data.
• Average differences between chatbot and Fitbit data were around 15 minutes; however, individual variations were significant, making prediction difficult.
• Participants reporting good sleep tended to estimate longer sleep durations than measured by Fitbit.

Conclusions:

• Chatbot-reported sleep times and Fitbit measurements show average similarity, but individual perception varies greatly.
• Subjective reporting of good sleep quality correlates with longer perceived sleep duration.
• Significant individual differences in sleep time perception underscore the limitations of relying solely on wearable devices for objective sleep assessment.