A continuous real-world dataset comprising wearable-based heart rate variability alongside sleep diaries

Area of Science:

• Physiology and Health Informatics
• Wearable Technology
• Biomedical Data Science

Background:

• Heart rate variability (HRV) monitoring offers insights into health conditions.
• Existing HRV datasets lack real-world complexity, being collected under controlled clinical settings.
• There is a need for robust, in-the-wild physiological data for health analytics.

Purpose of the Study:

• To collect and validate continuous, real-world physiological and motion data using smartwatches.
• To establish a benchmark dataset for in-the-wild HRV recordings.
• To support the development of predictive health analytics integrating wearable data and sleep patterns.

Main Methods:

• Collected continuous physiological and motion signals from 49 healthy adults over four weeks using smartwatches.
• Sampled recordings at 100ms intervals for short-term HRV computation from 5-minute segments.
• Validated data through signal frequency analysis, sensor-HRV correlation, and literature-based feature distribution checks.

Main Results:

• Successfully collected a novel dataset of continuous smartwatch recordings in a real-world setting.
• Validated the quality and reliability of the collected physiological and motion data.
• Demonstrated expected HRV and sleep-related feature distributions consistent with existing literature.

Conclusions:

• The developed dataset provides a valuable resource for benchmarking in-the-wild HRV.
• Enables future research in analyzing real-world physiological data for health insights.
• Supports advancements in predictive analytics for health monitoring using wearable technology and sleep data.