On an automatic delineator for arterial blood pressure waveforms

Feature Extraction Tool Using Temporal Landmarks in Arterial Blood Pressure and Photoplethysmography Waveforms

Arterial blood pressure (ABP) and photoplethysmography (PPG) waveforms both contain vital physiological information for the prevention and treatment of cardiovascular diseases. Extracted features from these waveforms have diverse clinical applications, including predicting hyper- and hypo-tension, estimating cardiac output from ABP, and monitoring blood pressure and nociception from PPG. However, the lack of standardized tools for feature extraction limits their exploration and clinical utilization. In this study, we propose an automatic feature extraction tool that first detects temporal location of landmarks within each cardiac cycle of ABP and PPG waveforms, including the systolic phase onset, systolic phase peak, dicrotic notch, and diastolic phase peak using the iterative envelope mean method. Then, based on these landmarks, extracts 852 features per cardiac cycle, encompassing time-, statistical-, and frequency-domains. The tool's ability to detect landmarks was evaluated using ABP and PPG waveforms from a large perioperative dataset (MLORD dataset) comprising 17,327 patients. We analyzed 34,267 cardiac cycles of ABP waveforms and 33,792 cardiac cycles of PPG waveforms. Additionally, to assess the tool's real-time landmark detection capability, we retrospectively analyzed 3,000 cardiac cycles of both ABP and PPG waveforms, collected from a Philips IntelliVue MX800 patient monitor. The tool's detection performance was assessed against markings by an experienced researcher, achieving average F1-scores and error rates for ABP and PPG as follows: (1) On MLORD dataset: systolic phase onset (99.77 %, 0.35 % and 99.52 %, 0.75 %), systolic phase peak (99.80 %, 0.30 % and 99.56 %, 0.70 %), dicrotic notch (98.24 %, 2.63 % and 98.72 %, 1.96 %), and diastolic phase peak (98.59 %, 2.11 % and 98.88 %, 1.73 %); (2) On real time data: systolic phase onset (98.18 %, 3.03 % and 97.94 %, 3.43 %), systolic phase peak (98.22 %, 2.97 % and 97.74 %, 3.77 %), dicrotic notch (97.72 %, 3.80 % and 98.16 %, 3.07 %), and diastolic phase peak (98.04 %, 3.27 % and 98.08 %, 3.20 %). This tool has significant potential for supporting clinical utilization of ABP and PPG waveform features and for facilitating feature-based machine learning models for various clinical applications where features derived from these waveforms play a critical role.

Artifact identification and removal methodologies for intracranial pressure signals: a systematic scoping review

Objective. Intracranial pressure measurement (ICP) is an essential component of deriving of multivariate data metrics foundational to improving understanding of high temporal relationships in cerebral physiology. A significant barrier to this work is artifact ridden data. As such, the objective of this review was to examine the existing literature pertinent to ICP artifact management.Methods.A search of five databases (BIOSIS, SCOPUS, EMBASE, PubMed, and Cochrane Library) was conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines with the PRISMA Extension for Scoping Review. The search question examined the methods for artifact management for ICP signals measured in human/animals.Results.The search yielded 5875 unique results. There were 19 articles included in this review based on inclusion/exclusion criteria and article references. Each method presented was categorized as: (1) valid ICP pulse detection algorithms and (2) ICP artifact identification and removal methods. Machine learning-based and filter-based methods indicated the best results for artifact management; however, it was not possible to elucidate a single most robust method.Conclusion.There is a significant lack of standardization in the metrics of effectiveness in artifact removal which makes comparison difficult across studies. Differences in artifacts observed on patient neuropathological health and recording methodologies have not been thoroughly examined and introduce additional uncertainty regarding effectiveness.Significance. This work provides critical insights into existing literature pertaining to ICP artifact management as it highlights holes in the literature that need to be adequately addressed in the establishment of robust artifact management methodologies.

Stand-alone segmentation of blood flow pulsatility measured with diffuse correlation spectroscopy

We present a stand-alone blood flow index (BFI) pulse segmentation method for diffuse correlation spectroscopy that uses a wavelet-based representation of the BFI signal at the cardiac frequency in place of an exogenous physiological reference. We use this wavelet-based segmentation method to quantify BFI waveform morphology in a cohort of 30 healthy adults. We demonstrate that the waveform morphology features obtained with the wavelet approach strongly agree with those obtained using an exogenous blood pressure reference signal. These results suggest the promise of stand-alone wavelet-based BFI segmentation for quantifying BFI waveform morphological features.

An algorithm to detect dicrotic notch in arterial blood pressure and photoplethysmography waveforms using the iterative envelope mean method

BACKGROUND AND OBJECTIVE

Detection of the dicrotic notch (DN) within a cardiac cycle is essential for assessment of cardiac output, calculation of pulse wave velocity, estimation of left ventricular ejection time, and supporting feature-based machine learning models for noninvasive blood pressure estimation, and hypotension, or hypertension prediction. In this study, we present a new algorithm based on the iterative envelope mean (IEM) method to detect automatically the DN in arterial blood pressure (ABP) and photoplethysmography (PPG) waveforms.

METHODS

The algorithm was evaluated on both ABP and PPG waveforms from a large perioperative dataset (MLORD dataset) comprising 17,327 patients. The analysis involved a total of 1,171,288 cardiac cycles for ABP waveforms and 3,424,975 cardiac cycles for PPG waveforms. To evaluate the algorithm's performance, the systolic phase duration (SPD) was employed, which represents the duration from the onset of the systolic phase to the DN in the cardiac cycle. Correlation plots and regression analysis were used to compare the algorithm against marked DN detection, while box plots and Bland-Altman plots were used to compare its performance with both marked DN detection and an established DN detection technique (second derivative). The marking of the DN temporal location was carried out by an experienced researcher using the help of the 'find_peaks' function from the scipy Python package, serving as a reference for the evaluation. The marking was visually validated by both an engineer and an anesthesiologist. The robustness of the algorithm was evaluated as the DN was made less visually distinct across signal-to-noise ratios (SNRs) ranging from -30 dB to -5 dB in both ABP and PPG waveforms.

RESULTS

The correlation between SPD estimated by the algorithm and that marked by the researcher is strong for both ABP (R2(87,343) =0.99, p<.001) and PPG (R2(86,764) =0.98, p<.001) waveforms. The algorithm had a lower mean error of DN detection (s): 0.0047 (0.0029) for ABP waveforms and 0.0046 (0.0029) for PPG waveforms, compared to 0.0693 (0.0770) for ABP and 0.0968 (0.0909) for PPG waveforms for the established 2nd derivative method. The algorithm has high rate of detectability of DN detection for SNR of >= -9 dB for ABP waveforms and >= -12 dB for PPG waveforms indicating robust performance in detecting the DN when it is less visibly distinct.

CONCLUSION

Our proposed IEM- based algorithm can detect DN in both ABP and PPG waveforms with low computational cost, even in cases where it is not distinctly defined within a cardiac cycle of the waveform ('DN-less signals'). The algorithm can potentially serve as a valuable, fast, and reliable tool for extracting features from ABP and PPG waveforms. It can be especially beneficial in medical applications where DN-based features, such as SPD, diastolic phase duration, and DN amplitude, play a significant role.

A signal processing tool for extracting features from arterial blood pressure and photoplethysmography waveforms

Arterial blood pressure (ABP) and photoplethysmography (PPG) waveforms contain valuable clinical information and play a crucial role in cardiovascular health monitoring, medical research, and managing medical conditions. The features extracted from PPG waveforms have various clinical applications ranging from blood pressure monitoring to nociception monitoring, while features from ABP waveforms can be used to calculate cardiac output and predict hypertension or hypotension. In recent years, many machine learning models have been proposed to utilize both PPG and ABP waveform features for these healthcare applications. However, the lack of standardized tools for extracting features from these waveforms could potentially affect their clinical effectiveness. In this paper, we propose an automatic signal processing tool for extracting features from ABP and PPG waveforms. Additionally, we generated a PPG feature library from a large perioperative dataset comprising 17,327 patients using the proposed tool. This PPG feature library can be used to explore the potential of these extracted features to develop machine learning models for non-invasive blood pressure estimation.

Measurement error of pulse pressure variation

Dynamic preload parameters are used to guide perioperative fluid management. However, reported cut-off values vary and the presence of a gray zone complicates clinical decision making. Measurement error, intrinsic to the calculation of pulse pressure variation (PPV) has not been studied but could contribute to this level of uncertainty. The purpose of this study was to quantify and compare measurement errors associated with PPV calculations. Hemodynamic data of patients undergoing liver transplantation were extracted from the open-access VitalDatabase. Three algorithms were applied to calculate PPV based on 1 min observation periods. For each method, different durations of sampling periods were assessed. Best Linear Unbiased Prediction was determined as the reference PPV-value for each observation period. A Bayesian model was used to determine bias and precision of each method and to simulate the uncertainty of measured PPV-values. All methods were associated with measurement error. The range of differential and proportional bias were [- 0.04%, 1.64%] and [0.92%, 1.17%] respectively. Heteroscedasticity influenced by sampling period was detected in all methods. This resulted in a predicted range of reference PPV-values for a measured PPV of 12% of [10.2%, 13.9%] and [10.3%, 15.1%] for two selected methods. The predicted range in reference PPV-value changes for a measured absolute change of 1% was [- 1.3%, 3.3%] and [- 1.9%, 4%] for these two methods. We showed that all methods that calculate PPV come with varying degrees of uncertainty. Accounting for bias and precision may have important implications for the interpretation of measured PPV-values or PPV-changes.

A signal processing tool for extracting features from arterial blood pressure and photoplethysmography waveforms

Arterial blood pressure (ABP) and photoplethysmography (PPG) waveforms contain valuable clinical information and play a crucial role in cardiovascular health monitoring, medical research, and managing medical conditions. The features extracted from PPG waveforms have various clinical applications ranging from blood pressure monitoring to nociception monitoring, while features from ABP waveforms can be used to calculate cardiac output and predict hypertension or hypotension. In recent years, many machine learning models have been proposed to utilize both PPG and ABP waveform features for these healthcare applications. However, the lack of standardized tools for extracting features from these waveforms could potentially affect their clinical effectiveness. In this paper, we propose an automatic signal processing tool for extracting features from ABP and PPG waveforms. Additionally, we generated a PPG feature library from a large perioperative dataset comprising 17,327 patients using the proposed tool. This PPG feature library can be used to explore the potential of these extracted features to develop machine learning models for non-invasive blood pressure estimation.

Uncertainties in the Analysis of Heart Rate Variability: A Systematic Review

Heart rate variability (HRV) is an important metric with a variety of applications in clinical situations such as cardiovascular diseases, diabetes mellitus, and mental health. HRV data can be potentially obtained from electrocardiography and photoplethysmography signals, then computational techniques such as signal filtering and data segmentation are used to process the sampled data for calculating HRV measures. However, uncertainties arising from data acquisition, computational models, and physiological factors can lead to degraded signal quality and affect HRV analysis. Therefore, it is crucial to address these uncertainties and develop advanced models for HRV analysis. Although several reviews of HRV analysis exist, they primarily focus on clinical applications, trends in HRV methods, or specific aspects of uncertainties such as measurement noise. This paper provides a comprehensive review of uncertainties in HRV analysis, quantifies their impacts, and outlines potential solutions. To the best of our knowledge, this is the first study that presents a holistic review of uncertainties in HRV methods and quantifies their impacts on HRV measures from an engineer's perspective. This review is essential for developing robust and reliable models, and could serve as a valuable future reference in the field, particularly for dealing with uncertainties in HRV analysis.

Pulse wave analysis as a tool for the evaluation of resuscitation therapy in septic shock

Objective. Pulse wave analysis (PWA) can provide insights into cardiovascular biomechanical properties. The use of PWA in critically ill patients, such as septic shock patients, is still limited, but it can provide complementary information on the cardiovascular effects of treatment when compared to standard indices outlined in international guidelines. Previous works have highlighted how sepsis induces severe cardiovascular derangement with altered arterial blood pressure waveform morphology and how resuscitation according to standard haemodynamic targets is not able to restore the physiological functioning of the cardiovascular system. The aim of this work is to test the effectiveness of PWA in characterizing arterial waveforms obtained from a swine experiment involving polymicrobial septic shock and resuscitation with different drugs.Methods. During the experiment, morphological aortic waveform features, such as indices related to the dicrotic notch and inflection point, were extracted by means of PWA techniques. Finally, all the PWA indices were used to compute a clustering classification (mini batch K-means) of the pigs according to the different phases of the experiment. This analysis aimed to test if PWA features alone could be used to distinguish between the different responses to the administered therapies.Results. The PWA indices highlighted different cardiovascular conditions of the pigs in response to different treatments, despite the mean haemodynamic values typically used to guide therapy administration being similar in all animals. The clustering algorithm was able to distinguish between the different phases of the experiment and the different responses of the animals based on the unique information derived from the aortic PWA.Conclusion. Even when used alone, PWA indices were highly informative when assessing therapy responses in cases of septic shock.Significance. A complex pathological condition like septic shock requires extensive monitoring without neglecting important information from commonly measured signals such as arterial blood pressure. Future studies are needed to understand how individual differences in the response to therapy are associated with different cardiovascular conditions that may become specific therapy targets.

Normative cerebral microvascular blood flow waveform morphology assessed with diffuse correlation spectroscopy

Microvascular cerebral blood flow exhibits pulsatility at the cardiac frequency that carries valuable information about cerebrovascular health. This study used diffuse correlation spectroscopy to quantify normative features of these waveforms in a cohort of thirty healthy adults. We demonstrate they are sensitive to changes in vascular tone, as indicated by pronounced morphological changes with hypercapnia. Further, we observe significant sex-based differences in waveform morphology, with females exhibiting higher flow, greater area-under-the-curve, and lower pulsatility. Finally, we quantify normative values for cerebral critical closing pressure, i.e., the minimum pressure required to maintain flow in a given vascular region.

Physiowise: A Physics-aware Approach to Dicrotic Notch Identification

Dicrotic Notch (DN), one of the most significant and indicative features of the arterial blood pressure (ABP) waveform, becomes less pronounced and thus harder to identify as a matter of aging and pathological vascular stiffness. Generalizable and automatic DN identification for such edge cases is even more challenging in the presence of unexpected ABP waveform deformations that happen due to internal and external noise sources or pathological conditions that cause hemodynamic instability. We propose a physics-aware approach, named Physiowise (PW), that first employs a cardiovascular model to augment the original ABP waveform and reduce unexpected deformations, then apply a set of predefined rules on the augmented signal to find DN locations. We have tested the proposed method on in-vivo data gathered from 14 pigs under hemorrhage and sepsis study. Our result indicates 52% overall mean error improvement with 16% higher detection accuracy within the lowest permitted error range of 30ms. An additional hybrid methodology is also proposed to allow combining augmentation with any application-specific user-defined rule set.

Associations between locus coeruleus MRI contrast and physiological responses to acute stress in younger and older adults

Acute stress activates the brain's locus coeruleus (LC)-noradrenaline system. Recent studies indicate that a magnetic resonance imaging (MRI)-based measure of LC structure is associated with better cognitive outcomes in later life. Yet despite the LC's documented role in promoting physiological arousal during acute stress, no studies have examined whether MRI-assessed LC structure is related to arousal responses to acute stress. In this study, 102 younger and 51 older adults completed an acute stress induction task while we assessed multiple measures of physiological arousal (heart rate, breathing rate, systolic and diastolic blood pressure, sympathetic tone, and heart rate variability, HRV). We used turbo spin echo MRI scans to quantify LC MRI contrast as a measure of LC structure. We applied univariate and multivariate approaches to assess how LC MRI contrast was associated with arousal at rest and during acute stress reactivity and recovery. In older participants, having higher caudal LC MRI contrast was associated with greater stress-related increases in systolic blood pressure and decreases in HRV, as well as lower HRV during recovery from acute stress. These results suggest that having higher caudal LC MRI contrast in older adulthood is associated with more pronounced physiological responses to acute stress. Further work is needed to confirm these patterns in larger samples of older adults.

Hemorrhagic risk prediction in coronary artery disease patients based on photoplethysmography and machine learning

Hemorrhagic events are the main focus of attention during antithrombosis therapy in patients with coronary artery disease (CAD). This study aims to investigate the potential of using photoplethysmography (PPG) and machine learning techniques to assess hemorrhagic risk in patients with CAD. A total of 1638 patients with CAD were enrolled from January 2018 to October 2019, among which 114 patients were observed to have at least one positive event. Importantly, 102 patients with 9933 records were finally retained for analysis in this study. Participants were required to collect data using the portable PPG acquisition device and the specially designed Android APP. The data was collected and uploaded to a remote server. Based on collected PPG signals, we extracted features in a total of 30 dimensions from time-domain, frequency-domain, and wavelet packet decomposition. Logistic regression, support vector regression, random forest, and XGBoost regression models were established to achieve hemorrhagic risk evaluation, and then, their performances were compared. In total, 10 features extracted from PPG showed statistical significance (p < 0.01) between negative and positive groups. The newly established XGBoost model performed best in the hemorrhagic risk evaluation experiment, wherein the mean area under the curve (AUC) with tenfold cross-validation was 0.762 ± 0.024 and the sensitivity and specificity were 0.679 ± 0.051 and 0.714 ± 0.014, respectively. We established a data acquisition system for PPG signal collection, and demonstrated that a set of features extracted from PPG and the proposed machine learning model are promising in the evaluation of hemorrhagic risk among patients with CAD. In comparison with the traditional HAS-BLED score, the proposed method can obtain the quantitative risk prediction probability from a single PPG record, which has the advantages of dynamics and continuity, and can provide timely feedback for doctors' antithrombotic treatment, which is of great significance for doctors to quickly determine the effectiveness of the treatment and adjust the timely treatment plans accordingly.

Detecting beats in the photoplethysmogram: benchmarking open-source algorithms

The photoplethysmogram (PPG) signal is widely used in pulse oximeters and smartwatches. A fundamental step in analysing the PPG is the detection of heartbeats. Several PPG beat detection algorithms have been proposed, although it is not clear which performs best.Objective:This study aimed to: (i) develop a framework with which to design and test PPG beat detectors; (ii) assess the performance of PPG beat detectors in different use cases; and (iii) investigate how their performance is affected by patient demographics and physiology.Approach:Fifteen beat detectors were assessed against electrocardiogram-derived heartbeats using data from eight datasets. Performance was assessed using theF1score, which combines sensitivity and positive predictive value.Main results:Eight beat detectors performed well in the absence of movement withF1scores of ≥90% on hospital data and wearable data collected at rest. Their performance was poorer during exercise withF1scores of 55%-91%; poorer in neonates than adults withF1scores of 84%-96% in neonates compared to 98%-99% in adults; and poorer in atrial fibrillation (AF) withF1scores of 92%-97% in AF compared to 99%-100% in normal sinus rhythm.Significance:Two PPG beat detectors denoted 'MSPTD' and 'qppg' performed best, with complementary performance characteristics. This evidence can be used to inform the choice of PPG beat detector algorithm. The algorithms, datasets, and assessment framework are freely available.

Subject-Based Model for Reconstructing Arterial Blood Pressure from Photoplethysmogram

The continuous prediction of arterial blood pressure (ABP) waveforms via non-invasive methods is of great significance for the prevention and treatment of cardiovascular disease. Photoplethysmography (PPG) can be used to reconstruct ABP signals due to having the same excitation source and high signal similarity. The existing methods of reconstructing ABP signals from PPG only focus on the similarities between systolic, diastolic, and mean arterial pressures without evaluating their global similarity. This paper proposes a deep learning model with a W-Net architecture to reconstruct ABP signals from PPG. The W-Net consists of two concatenated U-Net architectures, the first acting as an encoder and the second as a decoder to reconstruct ABP from PPG. Five hundred records of different lengths were used for training and testing. The experimental results yielded high values for the similarity measures between the reconstructed ABP signals and their reference ABP signals: the Pearson correlation, root mean square error, and normalized dynamic time warping distance were 0.995, 2.236 mmHg, and 0.612 mmHg on average, respectively. The mean absolute errors of the SBP and DBP were 2.602 mmHg and 1.450 mmHg on average, respectively. Therefore, the model can reconstruct ABP signals that are highly similar to the reference ABP signals.

COVID-19 Prediction With Machine Learning Technique From Extracted Features of Photoplethysmogram Morphology

At present, COVID-19 is spreading widely around the world. It causes many health problems, namely, respiratory failure and acute respiratory distress syndrome. Wearable devices have gained popularity by allowing remote COVID-19 detection, contact tracing, and monitoring. In this study, the correlation of photoplethysmogram (PPG) morphology between patients with COVID-19 infection and healthy subjects was investigated. Then, machine learning was used to classify the extracted features between 43 cases and 43 control subjects. The PPG data were collected from 86 subjects based on inclusion and exclusion criteria. The systolic-onset amplitude was 3.72% higher for the case group. However, the time interval of systolic-systolic was 7.69% shorter in the case than in control subjects. In addition, 12 out of 20 features exhibited a significant difference. The top three features included dicrotic-systolic time interval, onset-dicrotic amplitude, and systolic-onset time interval. Nine features extracted by heatmap based on the correlation matrix were fed to discriminant analysis, k-nearest neighbor, decision tree, support vector machine, and artificial neural network (ANN). The ANN showed the best performance with 95.45% accuracy, 100% sensitivity, and 90.91% specificity by using six input features. In this study, a COVID-19 prediction model was developed using multiple PPG features extracted using a low-cost pulse oximeter.

Analysis on Four Derivative Waveforms of Photoplethysmogram (PPG) for Fiducial Point Detection

Fiducial points of photoplethysmogram (PPG), first derivative PPG (VPG), and second derivative PPG (APG) are essential in extracting numerous parameters to diagnose cardiovascular disease. However, the fiducial points were usually detected using complex mathematical algorithms. Inflection points from derivatives waveforms are not thoroughly studied, whereas they can significantly assist in peak detection. This study is performed to investigate the derivative waveforms of PPG and use them to detect the important peaks of PPG, VPG, and APG. PPGs with different morphologies from 43 ischemic heart disease subjects are analyzed. Inflection points of the derivative waveforms up to the fourth level are observed, and consistent information (derivative markers) is used to detect the fiducial points of PPG, VPG, and APG with proper sequence. Moving average filter and simple thresholding techniques are applied to detect the primary points in VPG and the third derivative waveform. A total of twelve out of twenty derivative markers are found reliable in detecting fiducial points of two common types of PPG. Systolic peaks are accurately detected with 99.64% sensitivity and 99.38% positive predictivity using the 43 IHD dataset and Complex System Laboratory (CSL) Pulse Oximetry Artifact Labels database. The study has introduced the fourth derivative PPG waveform with four main points, which are significantly valuable for detecting the fiducial points of PPG, VPG, and APG.

Beyond the Baroreflex: A New Measure of Autonomic Regulation Based on the Time-Frequency Assessment of Variability, Phase Coherence and Couplings

For decades the role of autonomic regulation and the baroreflex in the generation of the respiratory sinus arrhythmia (RSA) - modulation of heart rate by the frequency of breathing - has been under dispute. We hypothesized that by using autonomic blockers we can reveal which oscillations and their interactions are suppressed, elucidating their involvement in RSA as well as in cardiovascular regulation more generally. R-R intervals, end tidal CO2, finger arterial pressure, and muscle sympathetic nerve activity (MSNA) were measured simultaneously in 7 subjects during saline, atropine and propranolol infusion. The measurements were repeated during spontaneous and fixed-frequency breathing, and apnea. The power spectra, phase coherence and couplings were calculated to characterise the variability and interactions within the cardiovascular system. Atropine reduced R-R interval variability (p < 0.05) in all three breathing conditions, reduced MSNA power during apnea and removed much of the significant coherence and couplings. Propranolol had smaller effect on the power of oscillations and did not change the number of significant interactions. Most notably, atropine reduced R-R interval power in the 0.145-0.6 Hz interval during apnea, which supports the hypothesis that the RSA is modulated by a mechanism other than the baroreflex. Atropine also reduced or made negative the phase shift between the systolic and diastolic pressure, indicating the cessation of baroreflex-dependent blood pressure variability. This result suggests that coherent respiratory oscillations in the blood pressure can be used for the non-invasive assessment of autonomic regulation.

Effects of using different algorithms and fiducial points for the detection of interbeat intervals, and different sampling rates on the assessment of pulse rate variability from photoplethysmography

OBJECTIVE

Pulse Rate Variability (PRV) has been widely used as a surrogate of Heart Rate Variability (HRV). However, there are several technical aspects that may affect the extraction of PRV information from pulse wave signals such as the photoplethysmogram (PPG). The aim of this study was to evaluate the effects of changing the algorithm and fiducial points used for determining inter-beat intervals (IBIs), as well as the PPG sampling rate, from simulated PPG signals with known PRV content.

METHODS

PPG signals were simulated using a proposed model, in which PRV information can be modelled. Two independent experiments were performed. First, 5 IBIs detection algorithms and 8 fiducial points were used for assessing PRV information from the simulated PPG signals, and time-domain and Poincaré plot indices were extracted and compared to the expected values according to the simulated PRV. The best combination of algorithms and fiducial points were determined for each index, using factorial designs. Then, using one of the best combinations, PPG signals were simulated with varying sampling rates. PRV indices were extracted and compared to the expected values using Student t-tests or Mann-Whitney U-tests.

RESULTS

From the first experiment, it was observed that AVNN and SD2 indices behaved similarly, and there was no significant influence of the fiducial points used. For other indices, there were several combinations that behaved similarly well, mostly based on the detection of the valleys of the PPG signal. There were differences according to the quality of the PPG signal. From the second experiment, it was observed that, for all indices but SDNN, the higher the sampling rate the better. AVNN and SD2 showed no statistical differences even at the lowest evaluated sampling rate (32 Hz), while RMSSD, pNN50, S, SD1 and SD1/SD2 showed good performance at sampling rates as low as 128 Hz.

CONCLUSION

The best combination of IBIs detection algorithms and fiducial points differs according to the application, but those based on the detection of the valleys of the PPG signal tend to show a better performance. The sampling rate of PPG signals for PRV analysis could be lowered to around 128 Hz, although it could be further lowered according to the application.

SIGNIFICANCE

The standardisation of PRV analysis could increase the reliability of this signal and allow for the comparison of results obtained from different studies. The obtained results allow for a first approach to establish guidelines for two important aspects in PRV analysis from PPG signals, i.e. the way the IBIs are segmented from PPG signals, and the sampling rate that should be used for these analyses. Moreover, a model for simulating PPG signals with PRV information has been proposed, which allows for the establishing of these guidelines while controlling for other variables, such as the quality of the PPG signal.

A Real-Time PPG Peak Detection Method for Accurate Determination of Heart Rate during Sinus Rhythm and Cardiac Arrhythmia

OBJECTIVE

We have developed a peak detection algorithm for accurate determination of heart rate, using photoplethysmographic (PPG) signals from a smartwatch, even in the presence of various cardiac rhythms, including normal sinus rhythm (NSR), premature atrial contraction (PAC), premature ventricle contraction (PVC), and atrial fibrillation (AF). Given the clinical need for accurate heart rate estimation in patients with AF, we developed a novel approach that reduces heart rate estimation errors when compared to peak detection algorithms designed for NSR.

METHODS

Our peak detection method is composed of a sequential series of algorithms that are combined to discriminate the various arrhythmias described above. Moreover, a novel Poincaré plot scheme is used to discriminate between basal heart rate AF and rapid ventricular response (RVR) AF, and to differentiate PAC/PVC from NSR and AF. Training of the algorithm was performed only with Samsung Simband smartwatch data, whereas independent testing data which had more samples than did the training data were obtained from Samsung's Gear S3 and Galaxy Watch 3.

RESULTS

The new PPG peak detection algorithm provides significantly lower average heart rate and interbeat interval beat-to-beat estimation errors-30% and 66% lower-and mean heart rate and mean interbeat interval estimation errors-60% and 77% lower-when compared to the best of the seven other traditional peak detection algorithms that are known to be accurate for NSR. Our new PPG peak detection algorithm was the overall best performers for other arrhythmias.

CONCLUSION

The proposed method for PPG peak detection automatically detects and discriminates between various arrhythmias among different waveforms of PPG data, delivers significantly lower heart rate estimation errors for participants with AF, and reduces the number of false negative peaks.

SIGNIFICANCE

By enabling accurate determination of heart rate despite the presence of AF with rapid ventricular response or PAC/PVCs, we enable clinicians to make more accurate recommendations for heart rate control from PPG data.

Effects of Contact Pressure in Reflectance Photoplethysmography in an In Vitro Tissue-Vessel Phantom

With the continued development and rapid growth of wearable technologies, PPG has become increasingly common in everyday consumer devices such as smartphones and watches. There is, however, minimal knowledge on the effect of the contact pressure exerted by the sensor device on the PPG signal and how it might affect its morphology and the parameters being calculated. This study explores a controlled in vitro study to investigate the effect of continually applied contact pressure on PPG signals (signal-to-noise ratio (SNR) and 17 morphological PPG features) from an artificial tissue-vessel phantom across a range of simulated blood pressure values. This experiment confirmed that for reflectance PPG signal measurements for a given anatomical model, there exists an optimum sensor contact pressure (between 35.1 mmHg and 48.1 mmHg). Statistical analysis shows that temporal morphological features are less affected by contact pressure, lending credit to the hypothesis that for some physiological parameters, such as heart rate and respiration rate, the contact pressure of the sensor is of little significance, whereas the amplitude and geometric features can show significant change, and care must be taken when using morphological analysis for parameters such as SpO2 and assessing autonomic responses.

A Delineator for Arterial Blood Pressure Waveform Analysis Based on a Deep Learning Technique

A deep learning technique based on semantic segmentation was implemented into the blood pressure detection points field. Two models were trained and evaluated in terms of a reference detector. The proposed methodology outperforms the reference detector in two of the three classic benchmarks and on signals from a public database that were modified with realistic test maneuvers and artifacts. Both models differentiate regions with valid information and artifacts. So far, no other delineator had shown this capacity.

Dicrotic Notch Identification: a Generalizable Hybrid Approach under Arterial Blood Pressure (ABP) Curve Deformations

Dicrotic Notch (DN) is a distinctive and clinically significant feature of the arterial blood pressure curve. Its automatic identification has been the focus of many kinds of research using either model-based or rule-based methodologies. However, since DN morphology is quite variant following the patient-specific underlying physiological and pathological conditions, its automatic identification with these methods is challenging. This work proposes a hybrid approach that employs both model-based and rule-based approaches to enhance DN detection's generalizability. We have tested our approach on ABP data gathered from 14 pigs. Our result strongly indicates 36% overall mean error improvement with maximum 52% and -11% accuracy enhancement and degradation in extreme cases.

Automatic Onsets and Systolic Peaks Detection and Segmentation of Arterial Blood Pressure Waveforms using Fully Convolutional Neural Networks

Arterial blood pressure (ABP) waveform is a common physiological signal that contains a wealth of cardiovascular information. According to the cardiac cycle, the ABP waveform is divided into rapid ejection, systolic and diastolic phases. Therefore, the characteristic points of the arterial blood pressure waveform, i.e. their onsets, systolic peaks, represent the timing of the minimum and maximum pressures. It is important to detect these characteristic points accurately. Recently, many researchers have introduced some feature points detection methods, but the accuracy is not particularly high. In this paper, a deep learning method is proposed to achieve periodic segmentation and feature points detection of ABP signals using a one-dimensional U-Net network. The network can split the ABP signal into two parts and accurately detect the feature points. The method is validated on an ABP dataset of 126 people, 500 people each. Performances are good at different tolerance thresholds, with an average time difference of less than 1.5 ms. Finally, the method performs with 99.79% and 99.79% sensitivity, 99.99% and 99.94% positive predictivity, and 0.23% and 0.27% error rates for both onsets and systolic peaks at a tolerance threshold of 30 ms. To our knowledge, this is the first paper to use deep learning methods for the onsets and systolic peaks detections of ABP signals.

Atrial Fibrillation Classification with Smart Wearables Using Short-Term Heart Rate Variability and Deep Convolutional Neural Networks

Atrial fibrillation (AF) is a type of cardiac arrhythmia affecting millions of people every year. This disease increases the likelihood of strokes, heart failure, and even death. While dedicated medical-grade electrocardiogram (ECG) devices can enable gold-standard analysis, these devices are expensive and require clinical settings. Recent advances in the capabilities of general-purpose smartphones and wearable technology equipped with photoplethysmography (PPG) sensors increase diagnostic accessibility for most populations. This work aims to develop a single model that can generalize AF classification across the modalities of ECG and PPG with a unified knowledge representation. This is enabled by approximating the transformation of signals obtained from low-cost wearable PPG sensors in terms of Pulse Rate Variability (PRV) to temporal Heart Rate Variability (HRV) features extracted from medical-grade ECG. This paper proposes a one-dimensional deep convolutional neural network that uses HRV-derived features for classifying 30-s heart rhythms as normal sinus rhythm or atrial fibrillation from both ECG and PPG-based sensors. The model is trained with three MIT-BIH ECG databases and is assessed on a dataset of unseen PPG signals acquired from wrist-worn wearable devices through transfer learning. The model achieved the aggregate binary classification performance measures of accuracy: 95.50%, sensitivity: 94.50%, and specificity: 96.00% across a five-fold cross-validation strategy on the ECG datasets. It also achieved 95.10% accuracy, 94.60% sensitivity, 95.20% specificity on an unseen PPG dataset. The results show considerable promise towards seamless adaptation of gold-standard ECG trained models for non-ambulatory AF detection with consumer wearable devices through HRV-based knowledge transfer.

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.

The underlying mechanism of intersite discrepancies in ejection time measurements from arterial waveforms and its validation in the Framingham Heart Study

Radial applanation tonometry is a well-established method for clinical hemodynamic assessment and is also becoming popular in wrist-worn fitness trackers. The time difference between the foot and the dicrotic notch of the arterial pressure waveform is a well-accepted approximation for the left ventricular ejection time (ET). However, several clinical studies have shown that ET measured from the radial pressure waveform deviates from that measured centrally. In this work, we consider the systolic wave and the dicrotic wave as two independent traveling waves and hypothesize that their wave speed difference leads to the intersite differences of measured ET (ΔET). Accordingly, we derived a mathematical dicrotic wave decomposition model and identified the most influential factors on ΔET via global sensitivity analysis. In our clinical validation on a heterogeneous cohort (N = 5,742) from the Framingham Heart Study (FHS), the local sensitivity analysis results resembled the sensitivity variation patterns of ΔET from model simulations. A regression analysis on FHS data, using morphological features of radial pressure waveforms to estimate the carotid ET, produced a root mean square error of 3.76 ms and R² of 0.91. The proposed dicrotic wave decomposition model can explain the intersite ET measurement discrepancies observed in the clinical data of FHS and can facilitate the precise identification of ET with radial pressure waveforms. Therefore, the proposed model will improve various physics-based pulse wave analysis methods as well as prospective artificial intelligence methods for tackling the subsequent big data produced from widespread wearable radial pressure monitoring.NEW & NOTEWORTHY Based on a new understanding of pressure wave propagation, we propose a novel dicrotic wave decomposition model considering the dicrotic wave as an independent traveling component. The proposed model can explain the mechanism underlying the intersite discrepancies in ejection time measurement from arterial waveforms and then, in principle, enhance the accuracy of both classical physics-based as well as more contemporary artificial intelligence-based pulse wave analysis methods in clinical and wearable radial blood pressure monitoring applications.

Differential effects of the blood pressure state on pulse rate variability and heart rate variability in critically ill patients

Heart rate variability (HRV) utilizes the electrocardiogram (ECG) and has been widely studied as a non-invasive indicator of cardiac autonomic activity. Pulse rate variability (PRV) utilizes photoplethysmography (PPG) and recently has been used as a surrogate for HRV. Several studies have found that PRV is not entirely valid as an estimation of HRV and that several physiological factors, including the pulse transit time (PTT) and blood pressure (BP) changes, may affect PRV differently than HRV. This study aimed to assess the relationship between PRV and HRV under different BP states: hypotension, normotension, and hypertension. Using the MIMIC III database, 5 min segments of PPG and ECG signals were used to extract PRV and HRV, respectively. Several time-domain, frequency-domain, and nonlinear indices were obtained from these signals. Bland–Altman analysis, correlation analysis, and Friedman rank sum tests were used to compare HRV and PRV in each state, and PRV and HRV indices were compared among BP states using Kruskal–Wallis tests. The findings indicated that there were differences between PRV and HRV, especially in short-term and nonlinear indices, and although PRV and HRV were altered in a similar manner when there was a change in BP, PRV seemed to be more sensitive to these changes.

Blood Pressure Morphology Assessment from Photoplethysmogram and Demographic Information Using Deep Learning with Attention Mechanism

Arterial blood pressure (ABP) is an important vital sign from which it can be extracted valuable information about the subject's health. After studying its morphology it is possible to diagnose cardiovascular diseases such as hypertension, so ABP routine control is recommended. The most common method of controlling ABP is the cuff-based method, from which it is obtained only the systolic and diastolic blood pressure (SBP and DBP, respectively). This paper proposes a cuff-free method to estimate the morphology of the average ABP pulse (ABPM¯) through a deep learning model based on a seq2seq architecture with attention mechanism. It only needs raw photoplethysmogram signals (PPG) from the finger and includes the capacity to integrate both categorical and continuous demographic information (DI). The experiments were performed on more than 1100 subjects from the MIMIC database for which their corresponding age and gender were consulted. Without allowing the use of data from the same subjects to train and test, the mean absolute errors (MAE) were 6.57 ± 0.20 and 14.39 ± 0.42 mmHg for DBP and SBP, respectively. For ABPM¯, R correlation coefficient and the MAE were 0.98 ± 0.001 and 8.89 ± 0.10 mmHg. In summary, this methodology is capable of transforming PPG into an ABP pulse, which obtains better results when DI of the subjects is used, potentially useful in times when wireless devices are becoming more popular.

Platform for Analysis and Labeling of Medical Time Series

Reliable and diverse labeled reference data are essential for the development of high-quality processing algorithms for medical signals, such as electrocardiogram (ECG) and photoplethysmogram (PPG). Here, we present the Platform for Analysis and Labeling of Medical time Series (PALMS) designed in Python. Its graphical user interface (GUI) facilitates three main types of manual annotations-(1) fiducials, e.g., R-peaks of ECG; (2) events with an adjustable duration, e.g., arrhythmic episodes; and (3) signal quality, e.g., data parts corrupted by motion artifacts. All annotations can be attributed to the same signal simultaneously in an ergonomic and user-friendly manner. Configuration for different data and annotation types is straightforward and flexible in order to use a wide range of data sources and to address many different use cases. Above all, configuration of PALMS allows plugging-in existing algorithms to display outcomes of automated processing, such as automatic R-peak detection, and to manually correct them where needed. This enables fast annotation and can be used to further improve algorithms. The GUI is currently complemented by ECG and PPG algorithms that detect characteristic points with high accuracy. The ECG algorithm reached 99% on the MIT/BIH arrhythmia database. The PPG algorithm was validated on two public databases with an F1-score above 98%. The GUI and optional algorithms result in an advanced software tool that allows the creation of diverse reference sets for existing datasets.

New algorithm to quantify cardiopulmonary interaction in patients with atrial fibrillation: a proof-of-concept study

BACKGROUND

Traditional formulas to calculate pulse pressure variation (PPV) cannot be used in patients with atrial fibrillation (AF). We have developed a new algorithm that accounts for arrhythmia-induced pulse pressure changes, allowing us to isolate and quantify ventilation-induced pulse pressure variation (VPPV). The robustness of the algorithm was tested in patients subjected to altered loading conditions. We investigated whether changes in VPPV imposed by passive leg raising (PLR) were proportional to the pre-PLR values.

METHODS

Consenting patients with active AF scheduled for an ablation of the pulmonary vein under general anaesthesia and mechanical ventilation were included. Loading conditions were altered by PLR. ECG and invasive pressure data were acquired during 60 s periods before and after PLR. A generalised additive model was constructed for each patient on each observation period. The impact of AF was modelled on the two preceding RR intervals of each beat (RR₀ and RR_-1). The impact of ventilation and the long-term pulse pressure trends were modelled as separate splines. Ventilation-induced pulse pressure variation was defined as the percentage of the maximal change in pulse pressure during the ventilation cycle.

RESULTS

Nine patients were studied. The predictive abilities of the models had a median r² of 0.92 (inter-quartile range: 89.2-94.2). Pre-PLR VPPV ranged from 0.1% to 27.9%. After PLR, VPPV decreased to 0-11.3% (P<0.014). The relation between the Pre-PLR values and the magnitude of the changes imposed by the PLR was statistically significant (P<0.001).

CONCLUSIONS

Our algorithm enables quantification of VPPV in patients with AF with the ability to detect changing loading conditions.

Prediction of Septic Shock Onset in ICU by Instantaneous Monitoring of Vital Signs

Septic Shock is a critical pathological state that affects patients entering the intensive care unit (ICU). Many studies have been directed to characterize and predict the onset of the septic shock, both in ICU and in the Emergency Department employing data extracted from the Electronic Health Records. Recently, machine learning algorithms have been successfully employed to help characterize septic shock in a more objective and automatic fashion. Only a few of these studies employ information contained in the continuously recorded vital signs such as electrocardiogram and arterial blood pressure. In particular, we have devised a novel feature estimation procedure able to consider instantaneous dynamics related to cardiovascular control. This work aims at developing a short-term prediction algorithm for identifying patients experiencing septic shock among a population of 100 septic patients extracted from the MIMIC-III clinical and waveform database. Among all the results obtained from several trained machine learning models, the best performance reached an AUC on the test set equal to 0.93 (Accuracy=0.85, Sensitivity=0.89 and Specificity=0.82).

Heart Rate Variability (HRV) and Pulse Rate Variability (PRV) for the Assessment of Autonomic Responses

Introduction: Heart Rate Variability (HRV) and Pulse Rate Variability (PRV), are non-invasive techniques for monitoring changes in the cardiac cycle. Both techniques have been used for assessing the autonomic activity. Although highly correlated in healthy subjects, differences in HRV and PRV have been observed under various physiological conditions. The reasons for their disparities in assessing the degree of autonomic activity remains unknown.

Methods: To investigate the differences between HRV and PRV, a whole-body cold exposure (CE) study was conducted on 20 healthy volunteers (11 male and 9 female, 30.3 ± 10.4 years old), where PRV indices were measured from red photoplethysmography signals acquired from central (ear canal, ear lobe) and peripheral sites (finger and toe), and HRV indices from the ECG signal. PRV and HRV indices were used to assess the effects of CE upon the autonomic control in peripheral and core vasculature, and on the relationship between HRV and PRV. The hypotheses underlying the experiment were that PRV from central vasculature is less affected by CE than PRV from the peripheries, and that PRV from peripheral and central vasculature differ with HRV to a different extent, especially during CE.

Results: Most of the PRV time-domain and Poincaré plot indices increased during cold exposure. Frequency-domain parameters also showed differences except for relative-power frequency-domain parameters, which remained unchanged. HRV-derived parameters showed a similar behavior but were less affected than PRV. When PRV and HRV parameters were compared, time-domain, absolute-power frequency-domain, and non-linear indices showed differences among stages from most of the locations. Bland-Altman analysis showed that the relationship between HRV and PRV was affected by CE, and that it recovered faster in the core vasculature after CE.

Conclusion: PRV responds to cold exposure differently to HRV, especially in peripheral sites such as the finger and the toe, and may have different information not available in HRV due to its non-localized nature. Hence, multi-site PRV shows promise for assessing the autonomic activity on different body locations and under different circumstances, which could allow for further understanding of the localized responses of the autonomic nervous system.

Adaptive Maximal Blood Flow Velocity Estimation From Transcranial Doppler Echos

OBJECTIVE

Novel applications of transcranial Doppler (TCD) ultrasonography, such as the assessment of cerebral vessel narrowing/occlusion or the non-invasive estimation of intracranial pressure (ICP), require high-quality maximal flow velocity waveforms. However, due to the low signal-to-noise ratio of TCD spectrograms, measuring the maximal flow velocity is challenging. In this work, we propose a calibration-free algorithm for estimating maximal flow velocities from TCD spectrograms and present a pertaining beat-by-beat signal quality index.

METHODS

Our algorithm performs multiple binary segmentations of the TCD spectrogram and then extracts the pertaining envelopes (maximal flow velocity waveforms) via an edge-following step that incorporates physiological constraints. The candidate maximal flow velocity waveform with the highest signal quality index is finally selected.

RESULTS

We evaluated the algorithm on 32 TCD recordings from the middle cerebral and internal carotid arteries in 6 healthy and 12 neurocritical care patients. Compared to manual spectrogram tracings, we obtained a relative error of -1.5%, when considering the whole waveform, and a relative error of -3.3% for the peak systolic velocity.

CONCLUSION

The feedback loop between the signal quality assessment and the binary segmentation yields a robust algorithm for maximal flow velocity estimation. Clinical Impact: The algorithm has already been used in our ICP estimation pipeline. By making the code and the data publicly available, we hope that the algorithm will be a useful building block for the development of novel TCD applications that require high-quality flow velocity waveforms.

Multiple Physiological Signals Fusion Techniques for Improving Heartbeat Detection: A Review

This paper presents a review of the techniques found in the literature that aim to achieve a robust heartbeat detection from fusing multi-modal physiological signals (e.g., electrocardiogram (ECG), blood pressure (BP), artificial blood pressure (ABP), stroke volume (SV), photoplethysmogram (PPG), electroencephalogram (EEG), electromyogram (EMG), and electrooculogram (EOG), among others). Techniques typically employ ECG, BP, and ABP, of which usage has been shown to obtain the best performance under challenging conditions. SV, PPG, EMG, EEG, and EOG signals can help increase performance when included within the fusion. Filtering, signal normalization, and resampling are common preprocessing steps. Delay correction between the heartbeats obtained over some of the physiological signals must also be considered, and signal-quality assessment to retain the best signal/s must be considered as well. Fusion is usually accomplished by exploiting regularities in the RR intervals; by selecting the most promising signal for the detection at every moment; by a voting process; or by performing simultaneous detection and fusion using Bayesian techniques, hidden Markov models, or neural networks. Based on the results of the review, guidelines to facilitate future comparison of the performance of the different proposals are given and promising future lines of research are pointed out.

Recent development of respiratory rate measurement technologies

Respiratory rate (RR) is an important physiological parameter whose abnormality has been regarded as an important indicator of serious illness. In order to make RR monitoring simple to perform, reliable and accurate, many different methods have been proposed for such automatic monitoring. According to the theory of respiratory rate extraction, methods are categorized into three modalities: extracting RR from other physiological signals, RR measurement based on respiratory movements, and RR measurement based on airflow. The merits and limitations of each method are highlighted and discussed. In addition, current works are summarized to suggest key directions for the development of future RR monitoring methodologies.

Use of zero-frequency resonator for automatically detecting systolic peaks of photoplethysmogram signal

This work investigates the application of zero-frequency resonator (ZFR) for detecting systolic peaks of photoplethysmogram (PPG) signals. Based on the authors' studies, they propose an automated noise-robust method, which consists of the central difference operation, the ZFR, the mean subtraction and averaging, the peak determination, and the peak rejection/acceptance rule. The method is evaluated using different kinds of PPG signals taken from the standard MIT-BIH polysomnographic database and Complex Systems Laboratory database and the recorded PPG signals at their Biomedical System Lab. The method achieves an average sensitivity (Se) of 99.95%, positive predictivity (Pp) of 99.89%, and overall accuracy (OA) of 99.84% on a total number of 116,673 true peaks. Evaluation results further demonstrate the robustness of the ZFR-based method for noisy PPG signals with a signal-to-noise ratio (SNR) ranging from 30 to 5 dB. The method achieves an average Se = 99.76%, Pp = 99.84%, and OA = 99.60% for noisy PPG signals with a SNR of 5 dB. Various results show that the method yields better detection rates for both noise-free and noisy PPG signals. The method is simple and reliable as compared with the complexity of signal processing techniques and detection performance of the existing detection methods.

CRSIDLab: A Toolbox for Multivariate Autonomic Nervous System Analysis Using Cardiorespiratory Identification

This paper presents the Cardiorespiratory System Identification Lab (CRSIDLab), a MATLAB-based software tool for multivariate autonomic nervous system (ANS) evaluation through heart rate variability (HRV) analysis and cardiorespiratory system identification. Based on a graphical user interface, CRSIDLab provides a complete set of tools including pre-processing cardiorespiratory data (electrocardiogram, continuous blood pressure, airflow, and instantaneous lung volume), power spectral density estimation, and multivariable cardiorespiratory system model identification. Parametrized multivariate models can assess both HRV and baroreflex sensitivity (BRS) by considering the causal relationship from respiration to heart rate (or its reciprocal, R-to-R interval - RRI) and from systolic blood pressure to RRI, for instance. The impulse response, estimated from the model, is used as a mathematical tool to effectively open the inherently closed-loop nature of the cardiorespiratory system, allowing the investigation of the dynamic response between pairs of cardiorespiratory variables. This system modeling approach provides information on gain and temporal behavior regarding dynamics, such as the baroreflex, complementing traditional HRV, and BRS indices. The toolbox is presented and used to investigate autonomic function in sleep apnea. The results show that, while traditional HRV indices were unable to differentiate between apneic and non-apneic subjects, the autonomic descriptors obtained from the multivariate system identification techniques were able to show vagal impairment in apneic compared to non-apneic subjects. Thus, CRSIDLab can help promote the use of cardiorespiratory system identification as a potentially more sensitive measure of ANS activity than classical HRV analysis.

Assessment of Signal Processing Methods for Measuring the Respiratory Rate in the Neonatal Intensive Care Unit

Knowledge of the pathological instabilities in the breathing pattern can provide valuable insights into the cardiorespiratory status of the critically-ill infant as well as their maturation level. This paper is concerned with the measurement of respiratory rate in premature infants. We compare the rates estimated from the chest impedance pneumogram, the ECG-derived respiratory rhythms, and the PPG-derived respiratory rhythms against those measured in the reference standard of breath detection provided by attending clinical staff during 165 manual breath counts. We demonstrate that accurate RR estimates can be produced from all sources for RR in the 40-80 bpm (breaths per min) range. We also conclude that the use of indirect methods based on the ECG or the PPG poses a fundamental challenge in this population due to their poor behavior at fast breathing rates (upward of 80 bpm).

Biomimetic pulsatile flows through flexible microfluidic conduits

We bring out unique aspects of the pulsatile flow of a blood analog fluid (Xanthan gum solution) in a biomimetic microfluidic channel. Pressure waveforms that mimic biologically consistent pulsations are applied on physiologically relevant cylindrical microchannels fabricated using polydimethylsiloxane. The in vivo features of the relevant waveforms like peak amplitude and dicrotic notch are reproduced in vitro. The deformation profiles exhibit viscoelastic behavior toward the end of each cycle. Further, the time-varying velocity profiles are critically analyzed. The local hydrodynamics within the microchannel is found to be more significantly affected by pressure waveform rather than the actual wall deformation and the velocity profile. These results are likely to bear far-reaching implications for assessing micro-circulatory dynamics in lab on a chip based microfluidic platforms that to a large extent replicate physiologically relevant conditions.

Hypertension Assessment via ECG and PPG Signals: An Evaluation Using MIMIC Database

Cardiovascular diseases (CVDs) have become the biggest threat to human health, and they are accelerated by hypertension. The best way to avoid the many complications of CVDs is to manage and prevent hypertension at an early stage. However, there are no symptoms at all for most types of hypertension, especially for prehypertension. The awareness and control rates of hypertension are extremely low. In this study, a novel hypertension management method based on arterial wave propagation theory and photoplethysmography (PPG) morphological theory was researched to explore the physiological changes in different blood pressure (BP) levels. Pulse Arrival Time (PAT) and photoplethysmogram (PPG) features were extracted from electrocardiogram (ECG) and PPG signals to represent the arterial wave propagation theory and PPG morphological theory, respectively. Three feature sets, one containing PAT only, one containing PPG features only, and one containing both PAT and PPG features, were used to classify the different BP categories, defined as normotension, prehypertension, and hypertension. PPG features were shown to classify BP categories more accurately than PAT. Furthermore, PAT and PPG combined features improved the BP classification performance. The F1 scores to classify normotension versus prehypertension reached 84.34%, the scores for normotension versus hypertension reached 94.84%, and the scores for normotension plus prehypertension versus hypertension reached 88.49%. This indicates that the simultaneous collection of ECG and PPG signals could detect hypertension.

The Role of Baroreflex Sensitivity in Acute Hypotensive Episodes Prediction in the Intensive Care Unit

A life threatening condition in Intensive Care Unit (ICU) is the Acute Hypotensive Episode (AHE). Patients experiencing an AHE may suffer from irreversible organ damage associated with increased mortality. Predicting the onset of AHE could be of pivotal importance to establish appropriate and timely interventions. We propose a method that, using waveforms widely acquired in ICU, like Arterial Blood Pressure (ABP) and Electrocardiogram (ECG), will extract features relative to the cardiac system to predict whether or not a patient will experience a hypotensive episode. Specifically, we want to assess if there are hidden patterns in the dynamics of baroreflex able to improve the prediction of AHEs. We will investigate the predictive power of features related to the baroreflex by performing classifications with and without them. Results are obtained using 17 classifiers belonging to different model families: classification trees, Support Vector Machines (SVMs), K-Nearest Neighbors (KNNs) replicated with different set of hyper-parameters and logistic regression. On average, the use of baroreflex features in the AHE prediction process increases the Area Under the Curve (AUC) by 10%.

Bayesian fusion of physiological measurements using a signal quality extension

OBJECTIVE

The fusion of multiple noisy labels for biomedical data (such as ECG annotations, which may be obtained from human experts or from automated systems) into a single robust annotation has many applications in physiologic monitoring. Directly modelling the difficulty of the task has the potential to improve the fusion of such labels. This paper proposes a means for the incorporation of task difficulty, as quantified by 'signal quality', into the fusion process.

APPROACH

We propose a Bayesian fusion model to infer a consensus through aggregating labels, where the labels are provided by multiple imperfect automated algorithms (or 'annotators'). Our model incorporates the signal quality of the underlying recording when fusing labels. We compare our proposed model with previously published approaches. Two publicly available datasets were used to demonstrate the feasibility of our proposed model: one focused on QT interval estimation in the ECG and the other focused on respiratory rate (RR) estimation from the photoplethysmogram (PPG). We inferred the hyperparameters of our model using maximum- a posteriori inference and Gibbs sampling.

MAIN RESULTS

For the QT dataset, our model significantly outperformed the previously published models (root-mean-square error of [Formula: see text] ms for our model versus [Formula: see text] ms from the best existing model) when fusing labels from only three annotators. For the RR dataset, no improvement was observed compared to the same model without signal quality modelling, where our model outperformed existing models (mean-absolute error of [Formula: see text] bpm for our model versus [Formula: see text] bpm from the best existing model). We conclude that our approach demonstrates the feasibility of using a signal quality metric as a confidence measure to improve label fusion.

SIGNIFICANCE

Our Bayesian learning model provides an extension over existing work to incorporate signal quality as a confidence measure to improve the reliability of fusing labels from biomedical datasets.

System dynamics of active and passive postural changes: Insights from principal dynamic modes analysis of baroreflex loop

The baroreflex being a key modulator of cardiovascular control ensures adequate blood pressure regulation under orthostatic stress which otherwise may cause severe hypotension. Contrary to conventional baroreflex sensitivity indices derived across a-priori traditional frequency bands, the present study is aimed at proposing new indices for the assessment of baroreflex drive which follows active (supine to stand-up) and passive (supine to head-up tilt) postural changes. To achieve this, a novel system identification approach of principal dynamic modes (PDM) was utilized to extract data-adaptive frequency components of closed-loop interactions between beat-to-beat interval and systolic blood pressure recorded from 10 healthy humans. We observed that the gain of low-pass global PDM of cardiac arm (:feedback reflex loop, mediated by pressure sensors to adjust heart rate in response to arterial blood pressure), and 0.2 Hz global PDM of mechanical arm (:feed-forward pathways, originating changes in arterial blood pressure in response to heart rate variations) may function as potential markers to distinguish active and passive orthostatic tests in healthy subjects.

Unsupervised Bayesian Inference to Fuse Biosignal Sensory Estimates for Personalizing Care

The role of sensing technologies, such as wearables, in delivering precision care is becoming widely acceptable. Given the very large quantities of sensor data that rapidly accumulate, there is a need to employ automated algorithms to label biosignal sensor data. In many real-life clinical applications, no such expert labels are available, and algorithms for processing sensor data must be relied upon, without access to the "ground truth." It is therefore extremely difficult to choose which algorithms to trust or discard at any point in time, where different algorithms may be optimal for different patients, or even for different points in time for the same patient. We propose two fully Bayesian approaches for fusing labels from independent and potentially correlated annotators (i.e., algorithms or, where available, experts). These are generative models to aggregate labels (i.e., the outputs of the algorithms, such as identified ECG morphology) in an unsupervised manner, to estimate jointly the assumed bias and precision of each algorithm without access to the ground truth. The latter fused estimate may then be used to infer the underlying ground truth. For the first time in the biomedical context, we show that modeling correlations between annotators, and fusing information concerning task difficulty (such as the estimated quality of the sensor data), improve these estimates with respect to commonly employed strategies in the literature. Also, we adopt a strongly Bayesian approach to inference using Gibbs sampling to improve estimates over the existing state of the art. We present results from applying the proposed pair of models to simulated and two publicly available biomedical datasets, to demonstrate proof-of-principle. We show that our proposed models outperform all existing approaches recreated from the literature. We also show that the proposed methods are robust when dealing with missing values (as often occurs in real-life biomedical applications), and that they are suitably efficient for use in real-time applications, thereby providing the basis for the reliable use of sensors for personalizing the care of the individual.