Autocorrelation Function Based Signal Quality Assessment on Photoplethysmographic Recordings for Opportunistic Accurate Estimation of RR Time Series

Interbeat Interval (IBI) from photoplethysmographic recordings (PPG) is usually employed as a surrogate measure of R-to-R (RR) time series measured using the electrocardiogram. This work focuses on the relationship of the IBI and RR time series agreement with the PPG quality using the PPG's autocorrelation function (ACF) as a signal quality assessment (SQA) index. We monitored 31 subjects for 25 minutes each by measuring PPG using two systems (Biopac PPG sensor sampled at 1 kHz (BS) and Empatica wristband measured PPG sampled at around 64 Hz (ES)) while the BS was also employed to measure ECG. This work also provides ways to synchronize both systems and correct some sampling frequency disparities before performing the agreement assessment and fair comparison between measuring systems. The agreement between IBI and RR for ACF SQA larger than 0.85 (good quality, (GQ)) is significantly different than for ACF SQA lower than 0.85 (fair quality (FQ)) for both systems (BS: 2.74 ms ± 1.18 ms (GQ) vs 3.68 ms ± 2.06 ms (FQ); p<8.2·10-5 and ES: 7.94 ms ± 8.03 ms (GQ) vs 14.1 ms ± 11.7 ms (FQ); p<2.6·106). Identifying PPG signal segments with high ACF SQA opportunistically provides IBI time series with better agreement.