Blood Pressure Estimation from Photoplethysmogram Using a Spectro-Temporal Deep Neural Network

Abstract

Blood pressure (BP) is a direct indicator of hypertension, a dangerous and potentially deadly condition. Regular monitoring of BP is thus important, but many people have aversion towards cuff-based devices, and their limitation is that they can only be used at rest. Using just a photoplethysmogram (PPG) to estimate BP is a potential solution investigated in our study. We analyzed the MIMIC III database for high-quality PPG and arterial BP waveforms, resulting in over 700 h of signals after preprocessing, belonging to 510 subjects. We then used the PPG alongside its first and second derivative as inputs into a novel spectro-temporal deep neural network with residual connections. We have shown in a leave-one-subject-out experiment that the network is able to model the dependency between PPG and BP, achieving mean absolute errors of 9.43 for systolic and 6.88 for diastolic BP. Additionally we have shown that personalization of models is important and substantially improves the results, while deriving a good general predictive model is difficult. We have made crucial parts of our study, especially the list of used subjects and our neural network code, publicly available, in an effort to provide a solid baseline and simplify potential comparison between future studies on an explicit MIMIC III subset.

Wearable Piezoelectric-Based System for Continuous Beat-to-Beat Blood Pressure Measurement

“…The voltage change of SBP and DBP feature points within adjacent PPW signals can be acquired, according to Equations (8) and (9). The pressure sensitivity of the piezoelectric sensor offers a unit conversion of 2 mmV/mmHg that implies that the pressure change is 0.5 mmHg per mV, according to Equation (10). Therefore, the pressure change of SBP and DBP feature points between the adjacent PPW signals were obtained by (11).…”

“…The experimental results indicated that the MAE ± SD for SBP and DBP were validated by a sphygmomanometer, and obtained 1.52 ± 9.45 mmHg and 0.39 ± 4.93 mmHg on 44 subjects. Slapniˇcar et al [10] created a blood pressure estimation model by PPG measurement and deep neural network. The results obtained MAE for SBP and DBP were, respectively, 9.43 and 6.88.…”

“…PTT is determined by the time delay between the R-peak of the ECG signal and the time point of the maximum slope of the PPG waveform [7,8]. The PTT can also be estimated from two-channel PPG signals recorded from two different locations to calculate the time delay between both waveforms [9,10]. Many studies presented several PTT-based mathematical models for blood pressure estimation by different models [11], including logarithmic model [12], linear model [13], inverse square model [14], and inverse model [15].…”

A Comparative Study of Brachial–Ankle Pulse Wave Velocity and Heart–Finger Pulse Wave Velocity in Korean Adults

“…In particular, ECG and PPG signals are widely used for the evaluation of cardiovascular function. Characteristics of PPG waveforms such as amplitude, peak-peak time, and dicrotic notch were used for blood pressure estimation [23][24][25][26][27], hypertension assessment [28], and cardiovascular risk evaluation [29]. Since noise-free signals are very important when using the feature points of a biosignal, silicon photomultipliers (SiPMs) were adopted instead of the conventional photodiode (PD) [30] to obtain clear PPG waveforms.…”

Blood pressure prediction by exploiting informative features from ICU patients’ ECG and PPG signals under a heterogeneous ensemble learning framework

“…In recent years, there has been some progress in the utilisation of deep learning techniques for the prediction of BP [29][30][31][32][33]. Su et al [30] devised a multi-layer recurrent neural network (RNN) network named DeepRNN and multi-task training strategy is utilized to train a model to predict systolic BP(SBP), diastolic BP(DBP) and mean BP(MBP) simultaneously.…”