Heart rate variability (HRV) is traditionally derived from RR interval time series of electrocardiography (ECG). Photoplethysmography (PPG) also reflects the cardiac rhythm since the mechanical activity of the heart is coupled to its electrical activity. Thus, theoretically, PPG can be used for determining the interval between successive heartbeats and heart rate variability. However, the PPG wave lags behind the ECG signal by the time required for transmission of pulse wave. In this study, finger-tip PPG and standard lead II ECG were recorded for five minutes from 10 healthy subjects at rest. The results showed a high correlation (median = 0.97) between the ECG-derived RR intervals and PPG-derived peak-to-peak (PP) intervals. PP variability was accurate (0.1 ms) as compared to RR variability. The time domain, frequency domain and Poincaré plot HRV parameters computed using RR interval method and PP interval method showed no significant differences (p < 0.05). The error analysis also showed insignificant differences between the HRV indices obtained by the two methods. Bland-Altman analysis showed high degree of agreement between the two methods for all the parameters of HRV. Thus, HRV can also be reliably estimated from the PPG based PP interval method.

Unobtrusive continuous monitoring of important vital signs and activity metrics has the potential to provide remote health monitoring, at-home screening, and rapid notification of critical events such as heart attacks, falls, or respiratory distress. This paper contains validation results of a wireless Bluetooth Low Energy (BLE) patch sensor consisting of two electrocardiography (ECG) electrodes, a microcontroller, a tri-axial accelerometer, and a BLE transceiver. The sensor measures heart rate, heart rate variability (HRV), respiratory rate, posture, steps, and falls and was evaluated on a total of 25 adult participants who performed breathing exercises, activities of daily living (ADLs), various stretches, stationary cycling, walking/running, and simulated falls. Compared to reference devices, the heart rate measurement had a mean absolute error (MAE) of less than 2 bpm, time-domain HRV measurements had an RMS error of less than 15 ms, respiratory rate had an MAE of 1.1 breaths per minute during metronome breathing, posture detection had an accuracy of over 95% in two of the three patch locations, steps were counted with an absolute error of less than 5%, and falls were detected with a sensitivity of 95.2% and specificity of 100%.

Motion and noise artifacts (MNA) have been a serious obstacle in realizing the potential of Photoplethysmogram (PPG) signals for real-time monitoring of vital signs. We present a statistical approach based on the computation of kurtosis and Shannon Entropy (SE) for the accurate detection of MNA in PPG data. The MNA detection algorithm was verified on multi-site PPG data collected from both laboratory and clinical settings. The accuracy of the fusion of kurtosis and SE metrics for the artifact detection was 99.0%, 94.8% and 93.3% in simultaneously recorded ear, finger and forehead PPGs obtained in a clinical setting, respectively. For laboratory PPG data recorded from a finger with contrived artifacts, the accuracy was 88.8%. It was identified that the measurements from the forehead PPG sensor contained the most artifacts followed by finger and ear. The proposed MNA algorithm can be implemented in real-time as the computation time was 0.14 seconds using Matlab®.

A time-frequency spectral method detected blood loss in spontaneously breathing subjects before the onset of significant changes in heart rate or blood pressure. Spectral amplitudes at the heart rate frequency band were found to significantly decrease during blood loss in spontaneously breathing subjects, whereas those at the breathing rate frequency band did not significantly change. This technique may serve as a valuable tool in intraoperative and trauma settings to detect and monitor hemorrhage.

Respiratory rate (RR) is an important vital sign marker of health, and it is often neglected due to a lack of unobtrusive sensors for objective and convenient measurement. The respiratory modulations present in simple photoplethysmogram (PPG) have been useful to derive RR using signal processing, waveform fiducial markers, and hand-crafted rules. An endto-end deep learning approach based on residual network (ResNet) architecture is proposed to estimate RR using PPG. This approach takes time-series PPG data as input, learns the rules through the training process that involved an additional synthetic PPG dataset generated to overcome the insufficient data problem of deep learning, and provides RR estimation as outputs. The inclusion of a synthetic dataset for training improved the performance of the deep learning model by 34%. The final mean absolute error performance of the deep learning approach for RR estimation was 2.5±0.6 brpm using 5-fold cross-validation in two widely used public PPG datasets (n=95) with reliable RR references. The deep learning model achieved comparable performance to that of a classical method, which was also implemented for comparison. With large real-world data and reference ground truth, deep learning can be valuable for RR or other vital sign monitoring using PPG and other physiological signals.