a A PPG signal with labelled features, adapted from (Kachuee et al. 2017). b Measured PPG signal (upper) and its second derivative (lower), indicating systolic and diastolic peaks, a-wave and b-wave. Adapted from (Elgendi 2012)

a A PPG signal with labelled features, adapted from (Kachuee et al. 2017). b Measured PPG signal (upper) and its second derivative (lower), indicating systolic and diastolic peaks, a-wave and b-wave. Adapted from (Elgendi 2012)

Source publication
Article
Full-text available
Conventional blood pressure (BP) measurement methods have different drawbacks such as being invasive, cuff-based or requiring manual operations. There is significant interest in the development of non-invasive, cuff-less and continual BP measurement based on physiological measurement. However, in these methods, extracting features from signals is c...

Similar publications

Conference Paper
Full-text available
Being electrocardiogram already an established method for analyzing cardiac health, it gained many researchers' interests to classify heartbeats accurately. In spite of having numerous works in this field, it still lacks obtaining high accuracy scores. In this paper, some well-known machine learning approaches are used by tuning and compared with o...
Article
Full-text available
Heart disease diagnosis is a very hard task in the medical field, so the mortality rate is increased every day. Also, the diagnosing process is implemented in recent times to predict heart disease. The method of diagnosing a disease in the medical field can be regarded not only as a new unknown situation to obtain clinical data and data collected f...

Citations

... Recently, researchers have shown increased interest in developing advanced models, such as Convolutional Neural Networks (CNN) [9][10][11][12][13]. Many of them tend to use automated feature extraction from raw PPG signals or images of PPG waveforms. ...
... The model introduced in [9] has a relatively high but acceptable standard deviation, particularly for SBP (5.59 ± 7.25) compared to other models. With a comparable number of subjects, the proposed CNN-LSTM model in [11] obtained 4.43 ± 6.09 for SBP and 3.23 ± 4.75 for DBP on 45 subjects. Whereas our model achieved a MAE ± SD of 2.03 ± 3.12 for SBP and 1.18 ± 1.70 for DBP on 40 subjects. ...
Article
Full-text available
Considerable research has been devoted to developing machine-learning models for continuous Blood Pressure (BP) estimation. A challenging problem that arises in this domain is the selection of optimal features with interpretable models for medical professionals. The aim of this study was to investigate evidence-based physiologically motivating features based on a solid physiological background of BP determinants. A powerful and compact set of features encompassing six physiologically oriented features was extracted in addition to another set of features consisting of six commonly used features for comparison purposes. In this study, we proposed a BP predictive model using Long Short-Term Memory (LSTM) networks with multi-stage transfer learning approach. The proposed model topology consists of three cascaded stages. First, a BP classification stage. Second, a Mean Arterial Pressure (MAP) regression stage to further approximate a quantity proportional to Vascular Resistance (VR) using the extracted Cardiac Output (CO) from the PPG signal. Third, the main BP estimation stage. The final stage (final BP prediction) is able to exploit embedded correlations between BP and the proposed features along with derived outputs carrying hemodynamic characteristics through the sub-sequence stages. We also constructed traditional single-stage Artificial Neural Network (ANN) and LSTM-based models to appraise the performance gain of our proposed model. The models were tested and evaluated on 40 subjects from the MIMIC II database. The LSTM-based multi-stage model attained a MAE ± SD of 2.03 ± 3.12 for SBP and 1.18 ± 1.70 mmHg for DBP. The proposed set of features resulted in drastic error reduction, of up to 86.21%, compared to models trained on the commonly used features. The superior performance of the proposed multi-stage model provides confirmatory evidence that the selected transferable features among the stages coupled with the high-performing multi-stage topology enhance blood pressure estimation accuracy using PPG signals. This indicates the compelling nature and sufficiency of the proposed efficient features set.
... The analysis of the second derivative of a PPG beat evaluates the ratios of the characteristic peaks. We included b/a as this is the most commonly used feature of the second derivative in BP estimation [42,53,54]. The second derivative is calculated from the reconstructed beat to reduce the impact of noise. ...
Article
Full-text available
Blood pressure (BP) is among the most important vital signals. Estimation of absolute BP solely using photoplethysmography (PPG) has gained immense attention over the last years. Available works differ in terms of used features as well as classifiers and bear large differences in their results. This work aims to provide a machine learning method for absolute BP estimation, its interpretation using computational methods and its critical appraisal in face of the current literature. We used data from three different sources including 273 subjects and 259,986 single beats. We extracted multiple features from PPG signals and its derivatives. BP was estimated by xgboost regression. For interpretation we used Shapley additive values (SHAP). Absolute systolic BP estimation using a strict separation of subjects yielded a mean absolute error of 9.456mmHg and correlation of 0.730. The results markedly improve if data separation is changed (MAE: 6.366mmHg, r: 0.874). Interpretation by means of SHAP revealed four features from PPG, its derivation and its decomposition to be most relevant. The presented approach depicts a general way to interpret multivariate prediction algorithms and reveals certain features to be valuable for absolute BP estimation. Our work underlines the considerable impact of data selection and of training/testing separation, which must be considered in detail when algorithms are to be compared. In order to make our work traceable, we have made all methods available to the public.
... ties, the principal drawback is the limited penetration of light into the tissue and their inability to capture haemodynamic parameters from arterial locations 27 . It has been proven that optical sensors can estimate heart rate from the skin surface and capillaries 25,26 ; however, the BP pulse wave does not reach capillaries effectively 28 , so BP cannot be captured from the capillaries ( Supplementary Fig. 1). ...
... The violin plot includes the box plot (defined as Q1 and Q3 quartiles, and median) with a kernel density estimation over the points. e, Comparison of the DBP estimation accuracy achieved with the GETs (Z-BP, green star) with Ag/AgCl gel-based BP 35 , photoplethysmography (PPG) 25 , ultrasound 52 , tonometry 53 , capacitive sensors 24 and various cuff-based methods (white circles) 54,55 . Only the works with relevant (m.e. and s.d.) data provided were included. ...
... 19 and 20), requiring only a primary calibration and subsequent continuous usage. Furthermore, our system demonstrates a capacity to run nocturnally with high fidelity without disturbing patients, which is not feasible with current obtrusive cuff-based monitors 3,50,51 or low-fidelity emerging cuffless BP monitoring solutions 25,26 owing to motion-noise artefacts that decrease the machine learning model efficiency 35 . Deployment of the system with miniaturized integrated circuits, wireless operation and data storage capabilities in the context of a smart-watch solution are among the future steps for translational research to develop a fully integrated wearable system. ...
Article
Full-text available
Continuous monitoring of arterial blood pressure (BP) in non-clinical (ambulatory) settings is essential for understanding numerous health conditions, including cardiovascular diseases. Besides their importance in medical diagnosis, ambulatory BP monitoring platforms can advance disease correlation with individual behaviour, daily habits and lifestyle, potentially enabling analysis of root causes, prognosis and disease prevention. Although conventional ambulatory BP devices exist, they are uncomfortable, bulky and intrusive. Here we introduce a wearable continuous BP monitoring platform that is based on electrical bioimpedance and leverages atomically thin, self-adhesive, lightweight and unobtrusive graphene electronic tattoos as human bioelectronic interfaces. The graphene electronic tattoos are used to monitor arterial BP for >300 min, a period tenfold longer than reported in previous studies. The BP is recorded continuously and non-invasively, with an accuracy of 0.2 ± 4.5 mm Hg for diastolic pressures and 0.2 ± 5.8 mm Hg for systolic pressures, a performance equivalent to Grade A classification. Self-adhesive bioimpedance graphene electronic tattoos enable accurate continuous blood pressure monitoring.
... In a word, monitoring theories of these methods limit their wide application in clinical and home use. Therefore, it is necessary to develop an easy-to-use and accurate method for continuous BP monitoring (Lázaro et al., 2019;Yang et al., 2021;Yen and Liao, 2022). ...
Article
Full-text available
In this article, a novel method for continuous blood pressure (BP) estimation based on multi-scale feature extraction by the neural network with multi-task learning (MST-net) has been proposed and evaluated. First, we preprocess the target (Electrocardiograph; Photoplethysmography) and label signals (arterial blood pressure), especially using peak-to-peak time limits of signals to eliminate the interference of the false peak. Then, we design a MST-net to extract multi-scale features related to BP, fully excavate and learn the relationship between multi-scale features and BP, and then estimate three BP values simultaneously. Finally, the performance of the developed neural network is verified by using a public multi-parameter intelligent monitoring waveform database. The results show that the mean absolute error ± standard deviation for systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial pressure (MAP) with the proposed method against reference are 4.04 ± 5.81, 2.29 ± 3.55, and 2.46 ± 3.58 mmHg, respectively; the correlation coefficients of SBP, DBP, and MAP are 0.96, 0.92, and 0.94, respectively, which meet the Association for the Advancement of Medical Instrumentation standard and reach A level of the British Hypertension Society standard. This study provides insights into the improvement of accuracy and efficiency of a continuous BP estimation method with a simple structure and without calibration. The proposed algorithm for BP estimation could potentially enable continuous BP monitoring by mobile health devices.
... According to the British Hypertension Society (BHS) standard, the results of their estimations achieve grade A for DBP and grade B for the mean arterial pressure (MAP). The authors of [14] developed a hybrid deep-learning-based model to predict BP from PPG and ECG signals. They implemented an automatic feature extraction layer in a deep-learning model to extract the optimal features from PPG signals for predicting BP. ...
Article
Full-text available
Blood pressure measurements are one of the most routinely performed medical tests globally. Blood pressure is an important metric since it provides information that can be used to diagnose several vascular diseases. Conventional blood pressure measurement systems use cuff-based devices to measure the blood pressure, which may be uncomfortable and sometimes burdensome to the subjects. Therefore, in this study, we propose a cuffless blood pressure estimation model based on Monte Carlo simulation (MCS). We propose a heterogeneous finger model for the MCS at wavelengths of 905 nm and 940 nm. After recording the photon intensities from the MCS over a certain range of blood pressure values, the actual photoplethysmography (PPG) signals were used to estimate blood pressure. We used both publicly available and self-made datasets to evaluate the performance of the proposed model. In case of the publicly available dataset for transmission-type MCS, the mean absolute errors are 3.32 ± 6.03 mmHg for systolic blood pressure (SBP), 2.02 ± 2.64 mmHg for diastolic blood pressure (DBP), and 1.76 ± 2.8 mmHg for mean arterial pressure (MAP). The self-made dataset is used for both transmission- and reflection-type MCSs; its mean absolute errors are 2.54 ± 4.24 mmHg for SBP, 1.49 ± 2.82 mmHg for DBP, and 1.51 ± 2.41 mmHg for MAP in the transmission-type case as well as 3.35 ± 5.06 mmHg for SBP, 2.07 ± 2.83 mmHg for DBP, and 2.12 ± 2.83 mmHg for MAP in the reflection-type case. The estimated results of the SBP and DBP satisfy the requirements of the Association for the Advancement of Medical Instrumentation (AAMI) standards and are within Grade A according to the British Hypertension Society (BHS) standards. These results show that the proposed model is efficient for estimating blood pressures using fingertip PPG signals.
Preprint
p>Cuffless blood pressure (BP) monitoring has gained great attention in the past twenty years considering its significant benefits in cardiovascular healthcare. However, the main challenge of this technology is the inaccurate BP modeling under activities, i.e., existing work have either been inappropriately validated with sufficient intra-individual BP variations, or did not show promising estimation accuracy under activities. In this study, a novel deep learning model UTransBPNet , featured in short- and long-range feature representation, is proposed aiming to improve the estimation accuracy and tracking capability of intra-individual BP changes under activities. The model performance was comprehensively evaluated in three different datasets, i.e., one public dataset (Dataset_MIMIC) and two datasets under daily activities (Dataset_Drink and Dataset_Exercise). Under subject-independent validation, the model achieved state-of-the-art performance in the Dataset_MIMIC, with the mean absolute differences (MADs) for systolic BP (SBP) and diastolic BP (DBP) of 4.38 and 2.25 mmHg, respectively. In addition, the model achieved strong tracking capability of intra-individual BP variations under activities, with the individual Pearson’ correlation coefficients for SBP and DBP of 0.61±0.17 and 0.62±0.13 (Dataset_Drink), 0.82±0.11 and 0.72±0.18 (Dataset_Exercise), respectively. Moreover, this study for the first time tested the generalization capability across different activities, and showed that with small-sized scenario-specific data for finetuning, our model showed good cross-scenario generalization capability which however degraded significantly when there are differences in the BP distribution and variation patterns between the datasets. In conclusion, UTransBPNet is generally a very promising deep learning model for accurate cuffless BP estimation and tracking BP changes. Future work should further investigate the influence of BP distribution and variation patterns on the generalization capability for building effective training dataset. </p
Article
Full-text available
Blood pressure (BP) estimation is one of the most popular and long-standing topics in health-care monitoring area. The utilization of machine learning (ML) and deep learning (DL) for BP prediction has made remarkable progress recently along with the development of ML and especially DL technologies, and the release of large-scale available datasets. In this survey, we present a comprehensive, systematic review about the recent advance of ML and DL for BP prediction. To start with, we systematically sort out the current progress from four perspectives. Then, we summarized commonly-used datasets, evaluation metrics as well as evaluation procedures (especially the usually ignored splitting strategy operation), which is followed by a critical analysis about the reported results. Next, we discussed several practical issues as well as newly-emerging techniques appeared in the research community of BP prediction. Also, we introduced the potential application of several advanced ML technologies in BP estimation. Last, we discussed the question of what a good BP estimator should look like?, and then a general proposal for an objective evaluation of model performance is given from the perspective of an ML researcher. Through this survey, we wish to provide a comprehensive, systematic, up-to-date (to Feb, 2022) review of related research on BP prediction using ML & DL methods, which may be helpful to researchers in this area. We also appeal an objective view of the progress reported in the relevant literatures in a more systematic manner. The experimental data & code and other useful resources are available at https://github.com/v3551G/BP-prediction-survey.
Article
A far cry from the bulky, uncomfortable cuff, the ultralight sensor takes measurements of the vital sign without the wearer feeling a thing.
Article
Estimating physiological parameters - such as blood pressure (BP) - from raw sensor data captured by noninvasive, wearable devices rely on either burdensome manual feature extraction designed by domain experts to identify key waveform characteristics and phases, or deep learning (DL) models that require extensive data collection. We propose the Data-Driven Guided Attention (DDGA) framework to optimize DL models to learn features supported by the underlying physiology and physics of the captured waveforms, with minimal expert annotation. With only a single template waveform cardiac cycle and its labelled fiducial points, we leverage dynamic time warping (DTW) to annotate all other training samples. DL models are trained to first identify them before estimating BP to inform them which regions of the input represent key phases of the cardiac cycle, yet we still grant the flexibility for DL to determine the optimal feature set from them. In this study, we evaluate DDGA's improvements to a BP estimation task for three prominent DL-based architectures with two datasets: 1) the MIMIC-III waveform dataset with ample training data and 2) a bio-impedance (Bio-Z) dataset with less than abundant training data. Experiments show that DDGA improves personalized BP estimation models by an average 8.14% in root mean square error (RMSE) when there is an imbalanced distribution of target values in a training set and improves model generalizability by an average 4.92% in RMSE when testing estimation of BP value ranges not previously seen in training.