Fast Quality-Aware AMDF Based Pulse Rate Estimation from Compressed PPG Measurements for Wearable Vital Signs Monitor.

Modern wearable or portable health monitoring devices are capable of photoplethysmogram (PPG) sensing, processing, analyzing, storing and transferring signal and parameters wirelessly but are generally energy constrained and have more false alarms under noisy PPG recordings. In this paper, we present computationally-efficient reliable pulse rate (PR) estimation in compressed sensing (CS) domain. The proposed CS-PPG based PR estimation method consists of measurement generation, high-pass filtering, average magnitude difference function (AMDF) features based signal quality assessment (SQA), and AMDF based quality-aware PR estimation. The proposed unified framework is evaluated using a wide variety of normal and pathological PPG signals taken from five standard databases. The proposed framework had an average sensitivity (SE) of 98.75% and specificity (SP) of 63.35%. Results show that the CS-PPG based quality-aware PR estimation method had a mean absolute error (MAE) of 3.1 ± 4.6%, Bland-Altman ratio (BAR) of 9.2% and root-mean-square error (RMSE) of 5.6 which are not only comparable with the results of the PR estimation method with the original PPG signals but also the proposed framework reduces 80% of the overall computational load. The proposed unified framework has great benefits in reducing processing time and energy consumption and thus can maximize battery lifetime of battery-operated health monitors.