A Supervised Approach to Robust Photoplethysmography Quality Assessment

SiamQuality: a ConvNet-based foundation model for photoplethysmography signals

Objective. Physiological data are often low quality and thereby compromises the effectiveness of related health monitoring. The primary goal of this study is to develop a robust foundation model that can effectively handle low-quality issue in physiological data.Approach. We introduce SiamQuality, a self-supervised learning approach using convolutional neural networks (CNNs) as the backbone. SiamQuality learns to generate similar representations for both high and low quality photoplethysmography (PPG) signals that originate from similar physiological states. We leveraged a substantial dataset of PPG signals from hospitalized intensive care patients, comprised of over 36 million 30 s PPG pairs.Main results. After pre-training the SiamQuality model, it was fine-tuned and tested on six PPG downstream tasks focusing on cardiovascular monitoring. Notably, in tasks such as respiratory rate estimation and atrial fibrillation detection, the model's performance exceeded the state-of-the-art by 75% and 5%, respectively. The results highlight the effectiveness of our model across all evaluated tasks, demonstrating significant improvements, especially in applications for heart monitoring on wearable devices.Significance. This study underscores the potential of CNNs as a robust backbone for foundation models tailored to physiological data, emphasizing their capability to maintain performance despite variations in data quality. The success of the SiamQuality model in handling real-world, variable-quality data opens new avenues for the development of more reliable and efficient healthcare monitoring technologies.

Pulse2AI: An Adaptive Framework to Standardize and Process Pulsatile Wearable Sensor Data for Clinical Applications

Goal: To establish Pulse2AI as a reproducible data preprocessing framework for pulsatile signals that generate high-quality machine-learning-ready datasets from raw wearable recordings. Methods: We proposed an end-to-end data preprocessing framework that adapts multiple pulsatile signal modalities and generates machine-learning-ready datasets agnostic to downstream medical tasks. Results: a dataset preprocessed by Pulse2AI improved systolic blood pressure estimation by 29.58%, from 11.41 to 8.03 mmHg in root-mean-square-error (RMSE) and its diastolic counterpart by 26.01%, from 7.93 to 5.87 mmHg in RMSE. For respiration rate (RR) estimation, Pulse2AI boosted performance by 19.69%, from 1.47 to 1.18 breaths per minute (BrPM) in mean-absolute-error (MAE). Conclusion: Pulse2AI turns pulsatile signals into machine learning (ML) ready datasets for arbitrary remote health monitoring tasks. We tested Pulse2AI on multiple pulsatile modalities and demonstrated its efficacy in two medical applications. This work bridges valuable assets in remote sensing and internet of medical things to ML-ready datasets for medical modeling.

Benchmarking of Sensor Configurations and Measurement Sites for Out-of-the-Lab Photoplethysmography

Photoplethysmography (PPG) is used for heart-rate monitoring in a variety of contexts and applications due to its versatility and simplicity. These applications, namely studies involving PPG data acquisition during day-to-day activities, require reliable and continuous measurements, which are often performed at the index finger or wrist. However, some PPG sensors are susceptible to saturation, motion artifacts, and discomfort upon their use. In this paper, an off-the-shelf PPG sensor was benchmarked and modified to improve signal saturation. Moreover, this paper explores the feasibility of using an optimized sensor in the lower limb as an alternative measurement site. Data were collected from 28 subjects with ages ranging from 18 to 59 years. To validate the sensors' performance, signal saturation and quality, wave morphology, performance of automatic systolic peak detection, and heart-rate estimation, were compared. For the upper and lower limb locations, the index finger and the first toe were used as reference locations, respectively. Lowering the amplification stage of the PPG sensor resulted in a significant reduction in signal saturation, from 18% to 0.5%. Systolic peak detection at rest using an automatic algorithm showed a sensitivity and precision of 0.99 each. The posterior wrist and upper arm showed pulse wave morphology correlations of 0.93 and 0.92, respectively. For these locations, peak detection sensitivity and precision were 0.95, 0.94 and 0.89, 0.89, respectively. Overall, the adjusted PPG sensors are a good alternative for obtaining high-quality signals at the fingertips, and for new measurement sites, the posterior pulse and the upper arm allow for high-quality signal extraction.

The 2023 wearable photoplethysmography roadmap

Photoplethysmography is a key sensing technology which is used in wearable devices such as smartwatches and fitness trackers. Currently, photoplethysmography sensors are used to monitor physiological parameters including heart rate and heart rhythm, and to track activities like sleep and exercise. Yet, wearable photoplethysmography has potential to provide much more information on health and wellbeing, which could inform clinical decision making. This Roadmap outlines directions for research and development to realise the full potential of wearable photoplethysmography. Experts discuss key topics within the areas of sensor design, signal processing, clinical applications, and research directions. Their perspectives provide valuable guidance to researchers developing wearable photoplethysmography technology.

PhysioKit: An Open-Source, Low-Cost Physiological Computing Toolkit for Single- and Multi-User Studies

The proliferation of physiological sensors opens new opportunities to explore interactions, conduct experiments and evaluate the user experience with continuous monitoring of bodily functions. Commercial devices, however, can be costly or limit access to raw waveform data, while low-cost sensors are efforts-intensive to setup. To address these challenges, we introduce PhysioKit, an open-source, low-cost physiological computing toolkit. PhysioKit provides a one-stop pipeline consisting of (i) a sensing and data acquisition layer that can be configured in a modular manner per research needs, and (ii) a software application layer that enables data acquisition, real-time visualization and machine learning (ML)-enabled signal quality assessment. This also supports basic visual biofeedback configurations and synchronized acquisition for co-located or remote multi-user settings. In a validation study with 16 participants, PhysioKit shows strong agreement with research-grade sensors on measuring heart rate and heart rate variability metrics data. Furthermore, we report usability survey results from 10 small-project teams (44 individual members in total) who used PhysioKit for 4-6 weeks, providing insights into its use cases and research benefits. Lastly, we discuss the extensibility and potential impact of the toolkit on the research community.

Variational Autoencoders for Biomedical Signal Morphology Clustering and Noise Detection

Accurate estimation of physiological biomarkers using raw waveform data from non-invasive wearable devices requires extensive data preprocessing. An automatic noise detection method in time-series data would offer significant utility for various domains. As data labeling is onerous, having a minimally supervised abnormality detection method for input data, as well as an estimation of the severity of the signal corruptness, is essential. We propose a model-free, time-series biomedical waveform noise detection framework using a Variational Autoencoder coupled with Gaussian Mixture Models, which can detect a range of waveform abnormalities without annotation, providing a confidence metric for each segment. Our technique operates on biomedical signals that exhibit periodicity of heart activities. This framework can be applied to any machine learning or deep learning model as an initial signal validator component. Moreover, the confidence score generated by the proposed framework can be incorporated into different models' optimization to construct confidence-aware modeling. We conduct experiments using dynamic time warping (DTW) distance of segments to validated cardiac cycle morphology. The result confirms that our approach removes noisy cardiac cycles and the remaining signals, classified as clean, exhibit a 59.92% reduction in the standard deviation of DTW distances. Using a dataset of bio-impedance data of 97885 cardiac cycles, we further demonstrate a significant improvement in the downstream task of cuffless blood pressure estimation, with an average reduction of 2.67 mmHg root mean square error (RMSE) of Diastolic Blood pressure and 2.13 mmHg RMSE of systolic blood pressure, with increases of average Pearson correlation of 0.28 and 0.08, with a statistically significant improvement of signal-to-noise ratio respectively in the presence of different synthetic noise sources. This enables burden-free validation of wearable sensor data for downstream biomedical applications.

[A method for photoplethysmography signal quality assessment fusing multi-class features with multi-scale series information]

Photoplethysmography (PPG) is often affected by interference, which could lead to incorrect judgment of physiological information. Therefore, performing a quality assessment before extracting physiological information is crucial. This paper proposed a new PPG signal quality assessment by fusing multi-class features with multi-scale series information to address the problems of traditional machine learning methods with low accuracy and deep learning methods requiring a large number of samples for training. The multi-class features were extracted to reduce the dependence on the number of samples, and the multi-scale series information was extracted by a multi-scale convolutional neural network and bidirectional long short-term memory to improve the accuracy. The proposed method obtained the highest accuracy of 94.21%. It showed the best performance in all sensitivity, specificity, precision, and F1-score metrics, compared with 6 quality assessment methods on 14 700 samples from 7 experiments. This paper provides a new method for quality assessment in small samples of PPG signals and quality information mining, which is expected to be used for accurate extraction and monitoring of clinical and daily PPG physiological information.

Automated Multi-Wavelength Quality Assessment of Photoplethysmography Signals Using Modulation Spectrum Shape Features

Photoplethysmography (PPG) is used to measure blood volume changes in the microvascular bed of tissue. Information about these changes along time can be used for estimation of various physiological parameters, such as heart rate variability, arterial stiffness, and blood pressure, to name a few. As a result, PPG has become a popular biological modality and is widely used in wearable health devices. However, accurate measurement of various physiological parameters requires good-quality PPG signals. Therefore, various signal quality indexes (SQIs) for PPG signals have been proposed. These metrics have usually been based on statistical, frequency, and/or template analyses. The modulation spectrogram representation, however, captures the second-order periodicities of a signal and has been shown to provide useful quality cues for electrocardiograms and speech signals. In this work, we propose a new PPG quality metric based on properties of the modulation spectrum. The proposed metric is tested using data collected from subjects while they performed various activity tasks contaminating the PPG signals. Experiments on this multi-wavelength PPG dataset show the combination of proposed and benchmark measures significantly outperforming several benchmark SQIs with improvements of 21.3% BACC (balanced accuracy) for green, 21.6% BACC for red, and 19.0% BACC for infrared wavelengths, respectively, for PPG quality detection tasks. The proposed metrics also generalize for cross-wavelength PPG quality detection tasks.

LSTM-based real-time signal quality assessment for blood volume pulse analysis

Remote photoplethysmogram (rPPG) is a low-cost method to extract blood volume pulse (BVP). Some crucial vital signs, such as heart rate (HR) and respiratory rate (RR) etc. can be achieved from BVP for clinical medicine and healthcare application. As compared to the conventional PPG methods, rPPG is more promising because of its non-contacted measurement. However, both BVP detection methods, especially rPPG, are susceptible to motion and illumination artifacts, which lead to inaccurate estimation of vital signs. Signal quality assessment (SQA) is a method to measure the quality of BVP signals and ensure the credibility of estimated physiological parameters. But the existing SQA methods are not suitable for real-time processing. In this paper, we proposed an end-to-end BVP signal quality evaluation method based on a long short-term memory network (LSTM-SQA). Two LSTM-SQA models were trained using the BVP signals obtained with PPG and rPPG techniques so that the quality of BVP signals derived from these two methods can be evaluated, respectively. As there is no publicly available rPPG dataset with quality annotations, we designed a training sample generation method with blind source separation, by which two kinds of training datasets respective to PPG and rPPG were built. Each dataset consists of 38400 high and low-quality BVP segments. The achieved models were verified on three public datasets (IIP-HCI dataset, UBFC-Phys dataset, and LGI-PPGI dataset). The experimental results show that the proposed LSTM-SQA models can effectively predict the quality of the BVP signal in real-time.

Channel Intensity and Edge-Based Estimation of Heart Rate via Smartphone Recordings

Smartphones, today, come equipped with a wide variety of sensors and high-speed processors that can capture, process, store, and communicate different types of data. Coupled with their ubiquity in recent years, these devices show potential as practical and portable healthcare monitors that are both cost-effective and accessible. To this end, this study focuses on examining the feasibility of smartphones in estimating the heart rate (HR), using video recordings of the users’ fingerprints. The proposed methodology involves two-stage processing that combines channel-intensity-based approaches (Channel-Intensity mode/Counter method) and a novel technique that relies on the spatial and temporal position of the recorded fingerprint edges (Edge-Detection mode). The dataset used here included 32 fingerprint video recordings taken from 6 subjects, using the rear camera of 2 smartphone models. Each video clip was first validated to determine whether it was suitable for Channel-Intensity mode or Edge-Detection mode, followed by further processing and heart rate estimation in the selected mode. The relative accuracy for recordings via the Edge-Detection mode was 93.04%, with a standard error of estimates (SEE) of 6.55 and Pearson’s correlation r > 0.91, while the Channel-Intensity mode showed a relative accuracy of 92.75%, with an SEE of 5.95 and a Pearson’s correlation r > 0.95. Further statistical analysis was also carried out using Pearson’s correlation test and the Bland–Altman method to verify the statistical significance of the results. The results thus show that the proposed methodology, through smartphones, is a potential alternative to existing technologies for monitoring a person’s heart rate.

Artifacts classification and apnea events detection in neck photoplethysmography signals

The novel pulse oximetry measurement site of the neck is a promising location for multi-modal physiological monitoring. Specifically, in the context of respiratory monitoring, in which it is important to have direct information about airflow. The neck makes this possible, in contrast to common photoplethysmography (PPG) sensing sites. However, this PPG signal is susceptible to artifacts that critically impair the signal quality. To fully exploit neck PPG for reliable physiological parameters extraction and apneas monitoring, this paper aims to develop two classification algorithms for artifacts and apnea detection. Features from the time, correlogram, and frequency domains were extracted. Two SVM classifiers with RBF kernels were trained for different window (W) lengths and thresholds (Thd) of corruption. For artifacts classification, the maximum performance was attained for the parameters combination of [W = 6s-Thd= 20%], with an average accuracy= 85.84%(ACC), sensitivity= 85.43%(SE) and specificity= 86.26%(SP). For apnea detection, the model [W = 10s-Thd= 50%] maximized all the performance metrics significantly (ACC= 88.25%, SE= 89.03%, SP= 87.42%). The findings of this proof of concept are significant for denoising novel neck PPG signals, and demonstrate, for the first time, that it is possible to promptly detect apnea events from neck PPG signals in an instantaneous manner. This could make a big impact in crucial real-time applications, like devices to prevent sudden-unexpected-death-in-epilepsy (SUDEP).

Impact of Label Noise on the Learning Based Models for a Binary Classification of Physiological Signal

Label noise is omnipresent in the annotations process and has an impact on supervised learning algorithms. This work focuses on the impact of label noise on the performance of learning models by examining the effect of random and class-dependent label noise on a binary classification task: quality assessment for photoplethysmography (PPG). PPG signal is used to detect physiological changes and its quality can have a significant impact on the subsequent tasks, which makes PPG quality assessment a particularly good target for examining the impact of label noise in the field of biomedicine. Random and class-dependent label noise was introduced separately into the training set to emulate the errors associated with fatigue and bias in labeling data samples. We also tested different representations of the PPG, including features defined by domain experts, 1D raw signal and 2D image. Three different classifiers are tested on the noisy training data, including support vector machine (SVM), XGBoost, 1D Resnet and 2D Resnet, which handle three representations, respectively. The results showed that the two deep learning models were more robust than the two traditional machine learning models for both the random and class-dependent label noise. From the representation perspective, the 2D image shows better robustness compared to the 1D raw signal. The logits from three classifiers are also analyzed, the predicted probabilities intend to be more dispersed when more label noise is introduced. From this work, we investigated various factors related to label noise, including representations, label noise type, and data imbalance, which can be a good guidebook for designing more robust methods for label noise in future work.

Optimized Signal Quality Assessment for Photoplethysmogram Signals Using Feature Selection

OBJECTIVE

With the increasing use of wearable healthcare devices for remote patient monitoring, reliable signal quality assessment (SQA) is required to ensure the high accuracy of interpretation and diagnosis on the recorded data from patients. Photoplethysmographic (PPG) signals non-invasively measured by wearable devices are extensively used to provide information about the cardiovascular system and its associated diseases. In this study, we propose an approach to optimize the quality assessment of the PPG signals.

METHODS

We used an ensemble-based feature selection scheme to enhance the prediction performance of the classification model to assess the quality of the PPG signals. Our approach for feature and subset size selection yielded the best-suited feature subset, which was optimized to differentiate between the clean and artifact corrupted PPG segments.

CONCLUSION

A high discriminatory power was achieved between two classes on the test data by the proposed feature selection approach, which led to strong performance on all dependent and independent test datasets. We achieved accuracy, sensitivity, and specificity rates of higher than 0.93, 0.89, and 0.97, respectively, for dependent test datasets, independent of heartbeat type, i.e., atrial fibrillation (AF) or non-AF data including normal sinus rhythm (NSR), premature atrial contraction (PAC), and premature ventricular contraction (PVC). For independent test datasets, accuracy, sensitivity, and specificity rates were greater than 0.93, 0.89, and 0.97, respectively, on PPG data recorded from AF and non-AF subjects. These results were found to be more accurate than those of all of the contemporary methods cited in this work.

SIGNIFICANCE

As the results illustrate, the advantage of our proposed scheme is its robustness against dynamic variations in the PPG signal during long-term 14-day recordings accompanied with different types of physical activities and a diverse range of fluctuations and waveforms caused by different individual hemodynamic characteristics, and various types of recording devices. This robustness instills confidence in the application of the algorithm to various kinds of wearable devices as a reliable PPG signal quality assessment approach.

Wrist Photoplethysmography Signal Quality Assessment for Reliable Heart Rate Estimate and Morphological Analysis

Photoplethysmographic (PPG) signals are mainly employed for heart rate estimation but are also fascinating candidates in the search for cardiovascular biomarkers. However, their high susceptibility to motion artifacts can lower their morphological quality and, hence, affect the reliability of the extracted information. Low reliability is particularly relevant when signals are recorded in a real-world context, during daily life activities. We aim to develop two classifiers to identify PPG pulses suitable for heart rate estimation (Basic-quality classifier) and morphological analysis (High-quality classifier). We collected wrist PPG data from 31 participants over a 24 h period. We defined four activity ranges based on accelerometer data and randomly selected an equal number of PPG pulses from each range to train and test the classifiers. Independent raters labeled the pulses into three quality levels. Nineteen features, including nine novel features, were extracted from PPG pulses and accelerometer signals. We conducted ten-fold cross-validation on the training set (70%) to optimize hyperparameters of five machine learning algorithms and a neural network, and the remaining 30% was used to test the algorithms. Performances were evaluated using the full features and a reduced set, obtained downstream of feature selection methods. Best performances for both Basic- and High-quality classifiers were achieved using a Support Vector Machine (Acc: 0.96 and 0.97, respectively). Both classifiers outperformed comparable state-of-the-art classifiers. Implementing automatic signal quality assessment methods is essential to improve the reliability of PPG parameters and broaden their applicability in a real-world context.

Design of a Realtime Photoplethysmogram Signal Quality Checker for Wearables and Edge Computing

Photoplethysmogram (PPG) signal is extensively used for deducing health parameters of patients in order to infer about physiological conditions of heart, blood pressure, respiratory patterns, and so on. Such analysis and estimations can be done accurately only on high quality PPG signals with very minimal artifacts. PPG signals collected from fitness grade and smart phone scenarios are prone to muscle artifacts and hence there is a need to assess the signal quality before using the signal. Although there are approaches available in the realm of machine learning and deep learning, they are computationally expensive and may not be suitable for a wearable or edge computing scenario. In this paper, we propose the design of a quality checker to check the quality of the signal that can be directly implemented on edge devices like smartwatch. The algorithm is tested on PPG data collected from wearable, ICU and medical grade devices. In the wearable scenario where the noise levels are very high, our algorithm has performed significantly better with a F_score of over 0.92. Further we show that by applying the proposed quality checker, the accuracy of the computed heart rate from a smart phone PPG-application significantly improves.

Statistical Description of SaO2–SpO2 Relationship for Model of Oxygenation in Premature Infants

A pulse oximeter model linking arterial (SaO2) and peripheral (SpO2) oxygen saturation is the terminal part of a mathematical model of neonatal oxygen transport. Previous studies have confirmed the overestimation of oxygen saturation measured by pulse oximetry in neonates compared to arterial oxygen saturation and the large variability of measured values over time caused by measurement inaccuracies. This work aimed to determine the SpO2 measurement noise that affects the biased SpO2 value at each time point and integrate the noise description with the systematic bias between SaO2 and SpO2. The SaO2–SpO2 bias was based on previously published clinical data from pathological patients younger than 60 days requiring ventilatory support. The statistical properties of the random SpO2 measurement noise were estimated from the SpO2 continuous recordings of 21 pathological and 21 physiological neonates. The result of the work is a comprehensive characterization of the properties of a pulse oximeter model describing the transfer of the input SaO2 value to the output SpO2 value, including the bias and noise typical for the bedside monitoring of neonates. These results will help to improve a computer model of neonatal oxygen transport.

Identification of Characteristic Points in Multivariate Physiological Signals by Sensor Fusion and Multi-Task Deep Networks

Identification of characteristic points in physiological signals, such as the peak of the R wave in the electrocardiogram and the peak of the systolic wave of the photopletismogram, is a fundamental step for the quantification of clinical parameters, such as the pulse transit time. In this work, we presented a novel neural architecture, called eMTUnet, to automate point identification in multivariate signals acquired with a chest-worn device. The eMTUnet consists of a single deep network capable of performing three tasks simultaneously: (i) localization in time of characteristic points (labeling task), (ii) evaluation of the quality of signals (classification task); (iii) estimation of the reliability of classification (reliability task). Preliminary results in overnight monitoring showcased the ability to detect characteristic points in the four signals with a recall index of about 1.00, 0.90, 0.90, and 0.80, respectively. The accuracy of the signal quality classification was about 0.90, on average over four different classes. The average confidence of the correctly classified signals, against the misclassifications, was 0.93 vs. 0.52, proving the worthiness of the confidence index, which may better qualify the point identification. From the achieved outcomes, we point out that high-quality segmentation and classification are both ensured, which brings the use of a multi-modal framework, composed of wearable sensors and artificial intelligence, incrementally closer to clinical translation.

Deep convolutional neural network-based signal quality assessment for photoplethysmogram

Quality assessment of bio-signals is important to prevent clinical misdiagnosis. With the introduction of mobile and wearable health care, it is becoming increasingly important to distinguish available signals from noise. The goal of this study was to develop a signal quality assessment technology for photoplethysmogram (PPG) widely used in wearable healthcare. In this study, we developed and verified a deep neural network (DNN)-based signal quality assessment model using about 1.6 million 5-s segment length PPG big data of about 29 GB from the MIMIC III PPG waveform database. The DNN model was implemented through a 1D convolutional neural network (CNN). The number of CNN layers, number of fully connected nodes, dropout rate, batch size, and learning rate of the model were optimized through Bayesian optimization. As a result, 6 CNN layers, 1,546 fully connected layer nodes, 825 batch size, 0.2 dropout rate, and 0.002 learning rate were needed for an optimal model. Performance metrics of the result of classifying waveform quality into 'Good' and 'Bad', the accuracy, specificity, sensitivity, area under the receiver operating curve, and area under the precision-recall curve were 0.978, 0.948, 0.993, 0.985, 0.980, and 0.969, respectively. Additionally, in the case of simulated real-time application, it was confirmed that the proposed signal quality score tracked the decrease in pulse quality well.

Hybrid Convolutional Networks for End-to-End Event Detection in Concurrent PPG and PCG Signals Affected by Motion Artifacts

The accurate detection of physiologically-related events in photopletismographic (PPG) and phocardiographic (PCG) signals, recorded by wearable sensors, is mandatory to perform the estimation of relevant cardiovascular parameters like the heart rate and the blood pressure. However, the measurement performed in uncontrolled conditions without clinical supervision leaves the detection quality particularly susceptible to noise and motion artifacts. The performed work proposed a new fully-automatic computational framework, based on convolutional networks, to identify and localize fiducial points in time as the foot, maximum slope and peak in PPG signal and the S1 sound in the PCG signal, both acquired by a custom chest sensor, described recently in the literature by our group. The novelty entailing a custom neural architecture to process sequentially the PPG and PCG signals. Tests were performed analysing four different acquisition conditions (rest, cycling, rest recovery and walking). Cross-validation results for the three PPG fiducial points showed identification accuracy greater than 93 % and localization error (RMSE) less than 10 ms. As expected, cycling and walking conditions provided worse results than rest and recovery, however reaching an accuracy greater than 90 % and a localization error lower than 15 ms. Likewise, the identification and localization error for S1 sound were greater than 90 % and lower than 25 ms. Overall, this study showcased the ability of the proposed technique to detect events with high accuracy not only for steady acquisitions but also during subject movements. We also showed that the proposed network outperformed traditional Shannon-energy-envelope method in the detection of S1 sound. Therefore, we argue that coupling chest sensors and deep learning processing techniques may disclose wearable devices to unobtrusively acquire health information, being less affected by noise and motion artifacts.

Predicting Prolonged Length of ICU Stay through Machine Learning

This study aimed to construct machine learning (ML) models for predicting prolonged length of stay (pLOS) in intensive care units (ICU) among general ICU patients. A multicenter database called eICU (Collaborative Research Database) was used for model derivation and internal validation, and the Medical Information Mart for Intensive Care (MIMIC) III database was used for external validation. We used four different ML methods (random forest, support vector machine, deep learning, and gradient boosting decision tree (GBDT)) to develop prediction models. The prediction performance of the four models were compared with the customized simplified acute physiology score (SAPS) II. The area under the receiver operation characteristic curve (AUROC), area under the precision-recall curve (AUPRC), estimated calibration index (ECI), and Brier score were used to measure performance. In internal validation, the GBDT model achieved the best overall performance (Brier score, 0.164), discrimination (AUROC, 0.742; AUPRC, 0.537), and calibration (ECI, 8.224). In external validation, the GBDT model also achieved the best overall performance (Brier score, 0.166), discrimination (AUROC, 0.747; AUPRC, 0.536), and calibration (ECI, 8.294). External validation showed that the calibration curve of the GBDT model was an optimal fit, and four ML models outperformed the customized SAPS II model. The GBDT-based pLOS-ICU prediction model had the best prediction performance among the five models on both internal and external datasets. Furthermore, it has the potential to assist ICU physicians to identify patients with pLOS-ICU risk and provide appropriate clinical interventions to improve patient outcomes.

Learning based Quality Indicator Aiding Heart Rate Estimation in Wrist-Worn PPG

Technological advancements and miniaturization of wearable sensors have enabled long-term pervasive physiological monitoring. Wrist-worn photoplethysmography (PPG) sensors, although quite popular owing to their form factor, suffer from poor signal quality in ambulatory settings due to motion artifacts. This affects the reliable estimation of vital cardiac parameters, especially during motion/activities of daily living. Hence, in this paper, we have developed a learningbased quality indicator engine (QIE), evaluating on 23 PPG records of the TROIKA database. The engine comprises the fundamental steps of frequency-domain feature extraction, feature selection and classification by an ensemble of decision trees, achieving an accuracy of 83% in the testing set. To the best of our knowledge, the proposed quality engine is the first to be evaluated on wrist-PPG data acquired during various physical activities and with respect to improvement in heart rate (HR) estimation. The QIE demonstrated an average improvement of 43% in HR estimation, when used in conjunction with state-ofthe-art WFPV algorithm.Clinical Relevance- The proposed quality indicator engine helps to increase the efficacy of vital parameter estimation (e.g. heart rate) from pervasive, wrist-worn PPG sensors on the backdrop of motion artifacts when used in ambulatory settings (e.g. activities of daily living).

Classification of blood pressure in critically ill patients using photoplethysmography and machine learning

OBJECTIVE

The aim of this study was to evaluate the capability of features extracted from photoplethysmography (PPG) based Pulse Rate Variability (PRV) to classify hypertensive, normotensive and hypotensive events, and to estimate mean arterial, systolic and diastolic blood pressure in critically ill patients.

METHODS

Time-domain, frequency-domain and non-linear indices from PRV were extracted from 5-min and 1-min segments obtained from PPG signals. These features were filtered using machine learning algorithms in order to obtain the optimal combination for the classification of hypertensive, hypotensive and normotensive events, and for the estimation of blood pressure.

RESULTS

5-min segments allowed for an improved performance in both classification and estimation tasks. Classification of blood pressure states showed around 70% accuracy and around 75% specificity. The sensitivity, precision and F1 scores were around 50%. In estimating mean arterial, systolic, and diastolic blood pressure, mean absolute errors as low as 2.55 ± 0.78 mmHg, 4.74 ± 2.33 mmHg, and 1.78 ± 0.14 mmHg were obtained, respectively. Bland-Altman analysis and Wilcoxon rank sum tests showed good agreement between real and estimated values, especially for mean and diastolic arterial blood pressures.

CONCLUSION

PRV-based features could be used for the classification of blood pressure states and the estimation of blood pressure values, although including additional features from the PPG waveform could improve the results.

SIGNIFICANCE

PRV contains information related to blood pressure, which may aid in the continuous, noninvasive, non-intrusive estimation of blood pressure and detection of hypertensive and hypotensive events in critically ill subjects.

How machine learning is impacting research in atrial fibrillation: implications for risk prediction and future management

There has been an exponential growth of artificial intelligence (AI) and machine learning (ML) publications aimed at advancing our understanding of atrial fibrillation (AF), which has been mainly driven by the confluence of two factors: the advances in deep neural networks (DeepNNs) and the availability of large, open access databases. It is observed that most of the attention has centred on applying ML for dvsetecting AF, particularly using electrocardiograms (ECGs) as the main data modality. Nearly a third of them used DeepNNs to minimize or eliminate the need for transforming the ECGs to extract features prior to ML modelling; however, we did not observe a significant advantage in following this approach. We also found a fraction of studies using other data modalities, and others centred in aims, such as risk prediction, AF management, and others. From the clinical perspective, AI/ML can help expand the utility of AF detection and risk prediction, especially for patients with additional comorbidities. The use of AI/ML for detection and risk prediction into applications and smart mobile health (mHealth) technology would enable ‘real time’ dynamic assessments. AI/ML could also adapt to treatment changes over time, as well as incident risk factors. Incorporation of a dynamic AI/ML model into mHealth technology would facilitate ‘real time’ assessment of stroke risk, facilitating mitigation of modifiable risk factors (e.g. blood pressure control). Overall, this would lead to an improvement in clinical care for patients with AF.

Explainability Metrics of Deep Convolutional Networks for Photoplethysmography Quality Assessment

Photoplethysmography (PPG) is a noninvasive way to monitor various aspects of the circulatory system, and is becoming more and more widespread in biomedical processing. Recently, deep learning methods for analyzing PPG have also become prevalent, achieving state of the art results on heart rate estimation, atrial fibrillation detection, and motion artifact identification. Consequently, a need for interpretable deep learning has arisen within the field of biomedical signal processing. In this paper, we pioneer novel explanatory metrics which leverage domain-expert knowledge to validate a deep learning model. We visualize model attention over a whole testset using saliency methods and compare it to human expert annotations. Congruence, our first metric, measures the proportion of model attention within expert-annotated regions. Our second metric, Annotation Classification, measures how much of the expert annotations our deep learning model pays attention to. Finally, we apply our metrics to compare between a signal based model and an image based model for PPG signal quality classification. Both models are deep convolutional networks based on the ResNet architectures. We show that our signal-based one dimensional model acts in a more explainable manner than our image based model; on average 50.78% of the one dimensional model's attention are within expert annotations, whereas 36.03% of the two dimensional model's attention are within expert annotations. Similarly, when thresholding the one dimensional model attention, one can more accurately predict if each pixel of the PPG is annotated as artifactual by an expert. Through this testcase, we demonstrate how our metrics can provide a quantitative and dataset-wide analysis of how explainable the model is.

Classification of Photoplethysmographic Signal Quality with Deep Convolution Neural Networks for Accurate Measurement of Cardiac Stroke Volume

As photoplethysmographic (PPG) signals are comprised of numerous pieces of important physiological information, they have been widely employed to measure many physiological parameters. However, only a high-quality PPG signal can provide a reliable physiological assessment. Unfortunately, PPG signals are easily corrupted by motion artifacts and baseline drift during recording. Although several rule-based algorithms have been developed for evaluating the quality of PPG signals, few artificial intelligence-based algorithms have been presented. Thus, this study aims to classify the quality of PPG signals by using two two-dimensional deep convolution neural networks (DCNN) when the PPG pulse is used to measure cardiac stroke volume (SV) by impedance cardiography. An image derived from a PPG pulse and its differential pulse is used as the input to the two DCNN models. To quantify the quality of individual PPG pulses, the error percentage of the beat-to-beat SV measured by our device and medis® CS 2000 synchronously is used to determine whether the pulse quality is high, middle, or low. Fourteen subjects were recruited, and a total of 3135 PPG pulses (1342 high quality, 73 middle quality, and 1720 low quality) were obtained. We used a traditional DCNN, VGG-19, and a residual DCNN, ResNet-50, to determine the quality levels of the PPG pulses. Their results were all better than the previous rule-based methods. The accuracies of VGG-19 and ResNet-50 were 0.895 and 0.925, respectively. Thus, the proposed DCNN may be applied for the classification of PPG quality and be helpful for improving the SV measurement in impedance cardiography.

Photoplethysmography based atrial fibrillation detection: a review

Atrial fibrillation (AF) is a cardiac rhythm disorder associated with increased morbidity and mortality. It is the leading risk factor for cardioembolic stroke and its early detection is crucial in both primary and secondary stroke prevention. Continuous monitoring of cardiac rhythm is today possible thanks to consumer-grade wearable devices, enabling transformative diagnostic and patient management tools. Such monitoring is possible using low-cost easy-to-implement optical sensors that today equip the majority of wearables. These sensors record blood volume variations-a technology known as photoplethysmography (PPG)-from which the heart rate and other physiological parameters can be extracted to inform about user activity, fitness, sleep, and health. Recently, new wearable devices were introduced as being capable of AF detection, evidenced by large prospective trials in some cases. Such devices would allow for early screening of AF and initiation of therapy to prevent stroke. This review is a summary of a body of work on AF detection using PPG. A thorough account of the signal processing, machine learning, and deep learning approaches used in these studies is presented, followed by a discussion of their limitations and challenges towards clinical applications.

Deep learning approaches for plethysmography signal quality assessment in the presence of atrial fibrillation

OBJECTIVE

Photoplethysmography (PPG) monitoring has been implemented in many portable and wearable devices we use daily for health and fitness tracking. Its simplicity and cost-effectiveness has enabled a variety of biomedical applications, such as continuous long-term monitoring of heart arrhythmias, fitness, and sleep tracking, and hydration monitoring. One major issue that can hinder PPG-based applications is movement artifacts, which can lead to false interpretations. In many implementations, noisy PPG signals are discarded. Misinterpreted or discarded PPG signals pose a problem in applications where the goal is to increase the yield of detecting physiological events, such as in the case of paroxysmal atrial fibrillation (AF)-a common episodic heart arrhythmia and a leading risk factor for stroke. In this work, we compared a traditional machine learning and deep learning approaches for PPG quality assessment in the presence of AF, in order to find the most robust method for PPG quality assessment.

APPROACH

The training data set was composed of 78 278 30 s long PPG recordings from 3764 patients using bedside patient monitors. Two different representations of PPG signals were employed-a time-series based (1D) one and an image-based (2D) one. Trained models were tested on an independent set of 2683 30 s PPG signals from 13 stroke patients.

MAIN RESULTS

ResNet18 showed a higher performance (0.985 accuracy, 0.979 specificity, and 0.988 sensitivity) than SVM and other deep learning approaches. 2D-based models were generally more accurate than 1D-based models.

SIGNIFICANCE

2D representation of PPG signal enhances the accuracy of PPG signal quality assessment.

Signal quality measure for pulsatile physiological signals using morphological features: Applications in reliability measure for pulse oximetry

Pulse oximetry is a noninvasive and low-cost physiological monitor that measures blood oxygen levels. While the noninvasive nature of pulse oximetry is advantageous, the estimates of oxygen saturation generated by these devices are prone to motion artifacts and ambient noise, reducing the reliability of such estimations. Clinicians combat this by assessing the quality of oxygen saturation estimation by visual inspection of the photoplethysmograph (PPG), which represents changes in pulsatile blood volume and is also generated by the pulse oximeter. In this paper, we propose six morphological features that can be used to determine the quality of the PPG signal and generate a signal quality index. Unlike many similar studies, this approach uses machine learning and does not require a separate signal, such as ECG, for reference. Multiple algorithms were tested against 46 30-min PPG segments of patients with cardiovascular and respiratory conditions, including atrial fibrillation, hypoxia, acute heart failure, pneumonia, ARDS, and pulmonary embolism. These signals were independently annotated for signal quality by two clinicians, with the union of their annotations used as the ground-truth. Similar to any physiological signal recorded in a clinical setting, the utilized dataset is also unbalanced in favor of good quality segments. The experiments showed that a cost-sensitive Support Vector Machine (SVM) outperformed other tested methods and was robust to the unbalanced nature of the data. Though the proposed algorithm was tested on PPG signals, the methodology remains agnostic to the dataset used, and may be applied to any type of pulsatile physiological signal.