Optimized Blood Pressure Classification by Features of Pulse Rate Variability and Asymmetry

Pulse-rate variability (PRV) is a rather interesting alternative in blood pressure (BP) estimation. Notwith-standing, the suitability of PRV for BP monitoring is under dispute, while the performance of the reported PRV studies could be improved. Five-minute electrocardiography (ECG) and PPG recordings of 202 patients from the MIMIC-II database were recruited and classified into nor-motensive (NT), prehypertensive (PHT) and hypertensive (HT). PRV and asymmetry analysis was performed using time-, frequency-domain and non-linear indices. HRV was used to verify the results using Bland-Altman (BA) and correlation. Multi-class (MCC) and single-class classification was performed with 10-fold cross-validation and a 20% test set. For all but NT group, correlation was high (ρ > 0.8, p < 0.05)for all features except for LF/HF. BA analysis suggests a high concordance between most PRV and HRV features (CI > 90%, BA ratio < 10%). MCC, NT and HT classification accuracy was up to 95%, 90% and 92.5% using 6-, 7- and 5-feature models, respectively. PRV is reliable in monitoring BP in critically-ill patients. Adding pulse-rate asymmetry to PRV analysis significantly improves the results and outperforms previous studies applying PRV for BP estimation using the same database.