Reliability of heart-rate-variability features derived from ultra-short ECG recordings and their validity in the assessment of cardiac autonomic neuropathy. (July 2021)

• Record Type: Journal Article
• Authors: Wehler, D.; Jelinek, H.F.; Gronau, A.; Wessel, N.; Kraemer, J.F.; Krones, R.; Penzel, T.
• Is Part Of: Biomedical signal processing and control. Volume 68(2021)
• Publication Date: 2021-07
• Subjects: Heart Rate Variability -- Cardiac autonomic neuropathy -- Ultra-short -- Reliability -- Diagnostics -- Electrocardiography; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2021.102651
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract: Objective: Our research aimed at providing methodologically strong evidence for the reliability of HRV analysis of ultra-short (10–150 s) ECGs compared to the benchmark of 300 s. Furthermore, we evaluated the retainment of HRV differences between participants with definite cardiac autonomic neuropathy (dCAN) and controls as ECG recording length was reduced to 10 s. Methods: Pre-processed ECG recordings from 11 individuals with dCAN and 76 healthy controls were used to determine SDNN, RMSSD, Low Frequency (LF) and High Frequency (HF) and assess whether the reliability of HRV features remains comparable for ultra-short ECG time series. For frequency domain analysis, the Lomb-Scargle-Periodogram (LSP) and Fast-Fourier-Transform (FFT) approaches were compared. Nonparametric areas under the curve (AUCs) were estimated to examine the overall ability of ultra-short HRV to classify dCAN from ECGs of 10–300 s. Results: RMSSD and HF were reliable and retained significant differences between control and dCAN from ECGs down to 10 s. SDNN and LF could be considered reliable in recordings of at least 30 s. LSP and FFT yielded comparable results. Most prevalent group differences were found for LF between 60 and 300 s. Conclusion: RMSSD and HF were reliable HRV features showing excellent potential for the classification of dCAN from 10-s ECG strips. For recordings longer than one minute, LF proved to be reliable and yielded best discrimination results. Significance: The current research indicated that ultra-short HRV analysis has the potential to become a time-efficient and accurate diagnostic tool for dCAN in clinical practice.