Fundamental Considerations of HRV Analysis in the Development of Real-Time Biofeedback Systems

Mariam Bahameish1 and Tony Stockman2
1PhD Student, 2Senior Lecturer


Abstract

Background: HRV biofeedback (HRVB) training is effective in improving physical and emotional health and resilience. However, there are challenges in delivering and interpreting biofeedback information. Objective: We developed a real-time HRVB system by exploring the minimum time window of RR-intervals that result in a reliable HRV analysis. Also, we investigated the appropriate HRV measures by examining changes between resting and deep breathing conditions and the trends across different time segments. Methods: We collected short-term HRV recordings using a PPG-based sensor device in three different conditions: at rest as a baseline measurement, during mentally stressed tasks based on the Trier Social Stress protocol, and during deep breathing as a post-stress recovery exercise. The RR recordings were then filtered, segmented and analysed. Each recording was divided into shorter time segments of 120, 60, 30, 20 and 10 seconds. To assess the reliability of ultra-short-term analysis (< 5 min), the HRV measures of each segment were compared to the standard 5 minutes as a benchmark using Pearson’s correlation test supported by Bland-Altman analysis. Also, the statistical test of the multilevel linear model and Tukey posthoc analysis were performed to evaluate the appropriate HRV measures. Results: According to Pearson’s correlation coefficient and Bland-Altman analysis, the shortest reliable time window for HRV analysis is 20-seconds in slow-paced breathing activities (rSDNN=.88, rLF=.83). This is an important design consideration because the RR signal should include at least one breathing cycle for robust frequency analysis. Regarding HRV features, the multilevel linear model revealed that SDNN, LF, LF/HF, and total power had significant changes during deep breathing compared with baseline measurements. All HRV measures demonstrated an excellent trend consistency following the Tukey posthoc analysis. Conclusions: The outcomes of this study contribute to designing an effective multimodal self-monitoring HRVB protocol, thus fostering the physiological and psychological well-being of individuals.