PSD Estimation in Artifact-Contaminated HRV Signals: A Methodological Comparison

Abstract

Heart rate variability (HRV) analysis commonly uses power spectral density (PSD) estimation and related band-based metrics, such as LF, HF, and LF/HF. However, the reliability of these metrics may change depending on data quality, recording duration, and the selected estimation method. In this study, the effects of ECG artifacts and recording duration on HRV spectral analysis were examined. R–R intervals obtained from synthetic ECG signals were used to compare different PSD methods. Performance analysis was also carried out under artifact conditions using multiple realizations. The results show that artifacts can affect R-peak detection and disturb the time structure of R–R intervals. They can also cause spectral distortions and deviations in LF/HF metrics. In addition, PSD results may change even in the reference condition depending on the method and duration, especially in short recordings. Therefore, data quality, recording duration, and method selection should be considered together for reliable HRV analysis.