Effect of missing RR-interval data on heart rate variability analysis in the time domain

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Time-domain analysis of RR-interval tachograms is crucial for assessing cardiac autonomic function.
• Missing data in electrocardiogram (ECG) recordings can significantly impact the reliability of heart rate variability (HRV) parameters.
• Understanding the robustness of different HRV metrics to data loss is essential for clinical interpretation.

Purpose of the Study:

• To evaluate the impact of simulated and real missing RR-interval data on commonly used time-domain HRV parameters.
• To identify which HRV parameters are most and least affected by data gaps.
• To provide insights into the reliability of HRV analysis in the presence of incomplete ECG data.

Main Methods:

• Simulated RR-interval data gaps (0-100s) were introduced into 2615 real RR-interval tachograms from the MIT-BIH normal sinus rhythm database.
• 1000 Monte Carlo runs were performed for each duration of missing data to calculate MeanNN, SDNN, SDSD, RMSSD, and pNN50.
• Relative errors were computed between original and incomplete tachograms; analysis was also performed on real missing data from sleep ECG.

Main Results:

• MeanNN demonstrated the highest robustness to missing RR-interval data, consistent with finite population correction (FPC) theory.
• pNN50 was found to be the most sensitive parameter to missing data.
• MeanNN also proved most robust with real-world missing data from capacitive-coupled ECG during sleep, showing similar parameter patterns to original data.

Conclusions:

• MeanNN is a reliable HRV parameter even with significant RR-interval data loss.
• pNN50 should be interpreted with caution when RR-interval data is incomplete.
• The findings support the use of MeanNN in clinical settings where ECG data quality may be compromised.