Effects of Missing Data on Heart Rate Variability Measured From A Smartwatch: Exploratory Observational Study.

Area of Science:

• Cardiovascular physiology
• Biomedical engineering
• Wearable technology

Background:

• Photoplethysmography (PPG) sensors in smartwatches are increasingly used for health monitoring.
• Missing data from PPG sensors can compromise the accuracy of heart rate variability (HRV) metrics.
• Robust methods are needed to handle missing data and validate PPG-derived HRV in real-world settings.

Purpose of the Study:

• To assess the impact of missing data on smartwatch-derived HRV metrics during rest and activity.
• To compare the agreement and consistency of HRV metrics between PPG and electrocardiogram (ECG) sensors.

Main Methods:

• Continuous, long-term data collection of interbeat intervals (IBI) using smartwatch PPG sensors.
• Simulated data degradation (10-60% missing data) to evaluate HRV metric stability.
• Comparison of HRV metrics (median IBI, STDRR, RMSDRR, LF, HF, LF/HF) between degraded and reference datasets.
• Validation of PPG-derived HRV against gold-standard chest-worn ECG.

Main Results:

• Median IBI remained stable up to 60% data loss at rest. STDRR and RMSDRR were stable up to 35% data loss at rest.
• During activity, STDRR remained stable up to 20% data loss, while median IBI and RMSDRR degraded at 10% missing data.
• Frequency-domain HRV metrics (LF, HF, LF/HF) were unstable with only 10% data loss.
• PPG showed moderate agreement (ICC=0.585) and consistency (ICC=0.589) for median IBI and moderate consistency (ICC=0.545) for LF power compared to ECG. Other metrics had poor agreement.

Conclusions:

• A methodology was developed for extracting stable HRV metrics from PPG data, minimizing data loss.
• Smartwatch PPG sensors are valuable for remote HRV monitoring, but require further research for optimal use in clinical settings.
• Best practices for utilizing PPG-derived HRV in medical contexts need to be established.