Search

BackgroundElectrocardiographic markers differentiating between death caused by ventricular arrhythmias and non-arrhythmic death could improve the selection of patients for implantable cardioverter-defibrillator (ICD) implantation. QRS fragmentation (fQRS) is a parameter of interest, but subject to debate. We investigated the association of an automatically quantified probability of fragmentation with the outcome in ICD patients.MethodsFrom a single-center retrospective registry, all patients implanted with an ICD between January 1996 and December 2018 were eligible for inclusion. Patients with active pacing were excluded. From the electronical medical record, clinical characteristics at implantation were collected and a 12-lead ECG was exported and analyzed by a previously validated machine-learning algorithm to quantify the probability of fQRS. To compare fQRS(+) and fQRS(−) patients, dichotomization was performed using the Youden index. Patients with a high probability of fragmentation in any region (anterior, inferior or lateral), were labeled fQRS(+). The impact of this fQRS probability on outcomes was investigated using Cox regression.ResultsA total of 1,242 patients with a mean age of 62.6 ± 11.5 years and a reduced left ventricular ejection fraction of 31 ± 12% were included of which 227 (18.3%) were female. The vast majority suffered from ischemic heart disease (64.3%) and were implanted in primary prevention (63.8%). 538 (43.3%) had a high probability of fragmentation in any region. Patients with a high probability of fragmentation had more frequently dilated cardiomyopathy (39.4% vs. 33.0%, p = 0.019), left bundle branch block (40.8% vs. 32.5%, p = 0.006) and a higher use of cardiac resynchronization therapy with defibrillator (CRT-D) devices (33.9% vs. 26.3%, p = 0.004). After adjustment in a multivariable Cox model, there was no significant association between the probability of global or regional fQRS and appropriate ICD therapy, inappropriate shock and short- or long-term mortality.ConclusionThere was no association between the automatically quantified probability of the presence of fQRS and outcome. This lack of predictive value might be due to the algorithm used, which identifies only the presence but not the severity of fragmentation.

Otago Exercise Program (OEP) is a rehabilitation program for older adults to improve frailty, sarcopenia, and balance. Accurate monitoring of patient involvement in OEP is challenging, as self-reports (diaries) are often unreliable. The development of wearable sensors and their use in Human Activity Recognition (HAR) systems has lead to a revolution in healthcare. However, the use of such HAR systems for OEP still shows limited performance. The objective of this study is to build an unobtrusive and accurate system to monitor OEP for older adults. Data was collected from 18 older adults wearing a single waist-mounted Inertial Measurement Unit (IMU). Two datasets were recorded, one in a laboratory setting, and one at the homes of the patients. A hierarchical system is proposed with two stages: 1) using a deep learning model to recognize whether the patients are performing OEP or activities of daily life (ADLs) using a 10-minute sliding window; 2) based on stage 1, using a 6-second sliding window to recognize the OEP sub-classes. Results showed that in stage 1, OEP could be recognized with window-wise f1-scores over 0.95 and Intersection-over-Union (IoU) f1-scores over 0.85 for both datasets. In stage 2, for the home scenario, four activities could be recognized with f1-scores over 0.8: ankle plantarflexors, abdominal muscles, knee bends, and sit-to-stand. These results showed the potential of monitoring the compliance of OEP using a single IMU in daily life. Also, some OEP sub-classes are possible to be recognized for further analysis.

Abstract Fragmented QRS (fQRS) is an electrocardiographic (ECG) marker of myocardial conduction abnormality, characterized by additional notches in the QRS complex. The presence of fQRS has been associated with an increased risk of all-cause mortality and arrhythmia in patients with cardiovascular disease. However, current binary visual analysis is prone to intra- and inter-observer variability and different definitions are problematic in clinical practice. Therefore, objective quantification of fQRS is needed and could further improve risk stratification of these patients. We present an automated method for fQRS detection and quantification. First, a novel robust QRS complex segmentation strategy is proposed, which combines multi-lead information and excludes abnormal heartbeats automatically. Afterwards extracted features, based on variational mode decomposition (VMD), phase-rectified signal averaging (PRSA) and the number of baseline-crossings of the ECG, were used to train a machine learning classifier (Support Vector Machine) to discriminate fragmented from non-fragmented ECG-traces using multi-center data and combining different fQRS criteria used in clinical settings. The best model was trained on the combination of two independent previously annotated datasets and, compared to these visual fQRS annotations, achieved Kappa scores of 0.68 and 0.44, respectively. We also show that the algorithm might be used in both regular sinus rhythm and irregular beats during atrial fibrillation. These results demonstrate that the proposed approach could be relevant for clinical practice by objectively assessing and quantifying fQRS. The study sets the path for further clinical application of the developed automated fQRS algorithm.

Abstract The ideal moment to withdraw respiratory supply of patients under Mechanical Ventilation at Intensive Care Units (ICU), is not easy to be determined for clinicians. Although the Spontaneous Breathing Trial (SBT) provides a measure of the patients’ readiness, there is still around 15–20% of predictive failure rate. This work is a proof of concept focused on adding new value to the prediction of the weaning outcome. Heart Rate Variability (HRV) and Cardiopulmonary Coupling (CPC) methods are evaluated as new complementary estimates to assess weaning readiness. The CPC is related to how the mechanisms regulating respiration and cardiac pumping are working simultaneously, and it is defined from HRV in combination with respiratory information. Three different techniques are used to estimate the CPC, including Time-Frequency Coherence, Dynamic Mutual Information and Orthogonal Subspace Projections. The cohort study includes 22 patients in pressure support ventilation, ready to undergo the SBT, analysed in the 24 h previous to the SBT. Of these, 13 had a successful weaning and 9 failed the SBT or needed reintubation –being both considered as failed weaning. Results illustrate that traditional variables such as heart rate, respiratory frequency, and the parameters derived from HRV do not differ in patients with successful or failed weaning. Results revealed that HRV parameters can vary considerably depending on the time at which they are measured. This fact could be attributed to circadian rhythms, having a strong influence on HRV values. On the contrary, significant statistical differences are found in the proposed CPC parameters when comparing the values of the two groups, and throughout the whole recordings. In addition, differences are greater at night, probably because patients with failed weaning might be experiencing more respiratory episodes, e.g. apneas during the night, which is directly related to a reduced respiratory sinus arrhythmia. Therefore, results suggest that the traditional measures could be used in combination with the proposed CPC biomarkers to improve weaning readiness.

Background An increase in beat‐to‐beat variability of repolarization (BVR) predicts arrhythmia onset in experimental models, but its clinical translation is not well established. We investigated the temporal changes in BVR before nonsustained ventricular tachycardia (nsVT) in patients with implantable cardioverter defibrillator (ICD). Methods and Results Patients with nsVT on 24‐hour Holter before ICD implantation for ischemic cardiomyopathy (ischemic cardiomyopathy+nsVT, n=43) or dilated cardiomyopathy (dilated cardiomyopathy+nsVT, n=37), matched ICD candidates without nsVT (ischemic cardiomyopathy‐nsVT, n=29 and dilated cardiomyopathy‐nsVT, n=26), and patients without ICD without structural heart disease (n=50) were studied. Digital Holter recordings from these patients were analyzed using a modified fiducial segment averaging technique to detect the QT interval. The nsVT episodes were semi‐automatically identified and QT‐BVR was assessed 1‐, 5‐, and 30‐minutes before nsVT, and at rest (at 3:00 am). Resting BVR was higher in ICD patients compared with controls without structural heart disease. In ICD patients with nsVT, BVR increased significantly 1‐minute pre‐nsVT in ischemic cardiomyopathy (2.21±0.59 ms, versus 5 minutes pre‐nsVT: 1.78±0.50 ms, P

Three-dimensional (3D) bio-printing has recently emerged as a crucial technology in tissue engineering, yet there are still challenges in selecting materials to obtain good print quality. Therefore, it is essential to study the influence of the chosen material (i.e., bio-ink) and the printing parameters on the final result. The “printability” of a bio-ink indicates its suitability for bio-printing. Hydrogels are a great choice because of their biocompatibility, but their printability is crucial for exploiting their properties and ensuring high printing accuracy. However, the printing settings are seldom addressed when printing hydrogels. In this context, this study explored the printability of double network (DN) hydrogels, from printing lines (1D structures) to lattices (2D structures) and 3D tubular structures, with a focus on printing accuracy. The DN hydrogel has two entangled cross-linked networks and a balanced mechanical performance combining high strength, toughness, and biocompatibility. The combination of poly (ethylene glycol)-diacrylate (PEDGA) and sodium alginate (SA) enables the qualities mentioned earlier to be met, as well as the use of UV to prevent filament collapse under gravity. Critical correlations between the printability and settings, such as velocity and viscosity of the ink, were identified. PEGDA/alginate-based double network hydrogels were explored and prepared, and printing conditions were improved to achieve 3D complex architectures, such as tubular structures. The DN solution ink was found to be unsuitable for extrudability; hence, glycerol was added to enhance the process. Different glycerol concentrations and flow rates were investigated. The solution containing 25% glycerol and a flow rate of 2 mm/s yielded the best printing accuracy. Thanks to these parameters, a line width of 1 mm and an angle printing inaccuracy of less than 1° were achieved, indicating good shape accuracy. Once the optimal parameters were identified, a tubular structure was achieved with a high printing accuracy. This study demonstrated a 3D printing hydrogel structure using a commercial 3D bio-printer (REGEMAT 3D BIO V1) by synchronizing all parameters, serving as a reference for future more complex 3D structures.

Wound management in Space is an important factor to be considered in future Human Space Exploration. It demands the development of reliable wound monitoring systems that will facilitate the assessment and proper care of wounds in isolated environments, such as Space. One possible system could be developed using liquid crystal films, which have been a promising solution for real-time in-situ temperature monitoring in healthcare, but they are not yet implemented in clinical practice. To progress in the latter, the goal of this study is twofold. First, it provides a full characterization of a sensing element composed of thermotropic liquid crystals arrays embedded between two elastomer layers, and second, it discusses how such a system compares against non-local infrared measurements. The sensing element evaluated here has an operating temperature range of 34–38°C, and a quick response time of approximately 0.25 s. The temperature distribution of surfaces obtained using this system was compared to the one obtained using the infrared thermography, a technique commonly used to measure temperature distributions at the wound site. This comparison was done on a mimicked wound, and results indicate that the proposed sensing element can reproduce the temperature distributions, similar to the ones obtained using infrared imaging. Although there is a long way to go before implementing the liquid crystal sensing element into clinical practice, the results of this work demonstrate that such sensors can be suitable for future wound monitoring systems.

Changes in respiratory rate have been found to be one of the early signs of health deterioration in patients. In remote environments where diagnostic tools and medical attention are scarce, such as deep space exploration, the monitoring of the respiratory signal becomes crucial to timely detect life-threatening conditions. Nowadays, this signal can be measured using wearable technology; however, the use of such technology is often hampered by the low quality of the recordings, which leads more often to wrong diagnosis and conclusions. Therefore, to apply these data in diagnosis analysis, it is important to determine which parts of the signal are of sufficient quality. In this context, this study aims to evaluate the performance of a signal quality assessment framework, where two machine learning algorithms (support vector machine–SVM, and convolutional neural network–CNN) were used. The models were pre-trained using data of patients suffering from chronic obstructive pulmonary disease. The generalization capability of the models was evaluated by testing them on data from a different patient population, presenting normal and pathological breathing. The new patients underwent bariatric surgery and performed a controlled breathing protocol, displaying six different breathing patterns. Data augmentation (DA) and transfer learning (TL) were used to increase the size of the training set and to optimize the models for the new dataset. The effect of the different breathing patterns on the performance of the classifiers was also studied. The SVM did not improve when using DA, however, when using TL, the performance improved significantly (p < 0.05) compared to DA. The opposite effect was observed for CNN, where the biggest improvement was obtained using DA, while TL did not show a significant change. The models presented a low performance for shallow, slow and fast breathing patterns. These results suggest that it is possible to classify respiratory signals obtained with wearable technologies using pre-trained machine learning models. This will allow focusing on the relevant data and avoid misleading conclusions because of the noise, when designing bio-monitoring systems.

Objectives: Sleep time information is essential for monitoring of obstructive sleep apnea (OSA), as the severity assessment depends on the number of breathing disturbances per hour of sleep. However, clinical procedures for sleep monitoring rely on numerous uncomfortable sensors, which could affect sleeping patterns. Therefore, an automated method to identify sleep intervals from unobtrusive data is required. However, most unobtrusive sensors suffer from data loss and sensitivity to movement artifacts. Thus, current sleep detection methods are inadequate, as these require long intervals of good quality. Moreover, sleep monitoring of OSA patients is often less reliable due to heart rate disturbances, movement and sleep fragmentation. The primary aim was to develop a sleep-wake classifier for sleep time estimation of suspected OSA patients, based on single short-term segments of their cardiac and respiratory signals. The secondary aim was to define metrics to detect OSA patients directly from their predicted sleep-wake pattern and prioritize them for clinical diagnosis.Methods: This study used a dataset of 183 suspected OSA patients, of which 36 test subjects. First, a convolutional neural network was designed for sleep-wake classification based on healthier patients (AHI < 10). It employed single 30 s epochs of electrocardiograms and respiratory inductance plethysmograms. Sleep information and Total Sleep Time (TST) was derived for all patients using the short-term segments. Next, OSA patients were detected based on the average confidence of sleep predictions and the percentage of sleep-wake transitions in the predicted sleep architecture.Results: Sleep-wake classification on healthy, mild and moderate patients resulted in moderate κ scores of 0.51, 0.49, and 0.48, respectively. However, TST estimates decreased in accuracy with increasing AHI. Nevertheless, severe patients were detected with a sensitivity of 78% and specificity of 89%, and prioritized for clinical diagnosis. As such, their inaccurate TST estimate becomes irrelevant. Excluding detected OSA patients resulted in an overall estimated TST with a mean bias error of 21.9 (± 55.7) min and Pearson correlation of 0.74 to the reference.Conclusion: The presented framework offered a realistic tool for unobtrusive sleep monitoring of suspected OSA patients. Moreover, it enabled fast prioritization of severe patients for clinical diagnosis.

Feature selection techniques are very useful approaches for dimensionality reduction in data analysis. They provide interpretable results by reducing the dimensions of the data to a subset of the original set of features. When the data lack annotations, unsupervised feature selectors are required for their analysis. Several algorithms for this aim exist in the literature, but despite their large applicability, they can be very inaccessible or cumbersome to use, mainly due to the need for tuning non-intuitive parameters and the high computational demands. In this work, a publicly available ready-to-use unsupervised feature selector is proposed, with comparable results to the state-of-the-art at a much lower computational cost. The suggested approach belongs to the methods known as spectral feature selectors. These methods generally consist of two stages: manifold learning and subset selection. In the first stage, the underlying structures in the high-dimensional data are extracted, while in the second stage a subset of the features is selected to replicate these structures. This paper suggests two contributions to this field, related to each of the stages involved. In the manifold learning stage, the effect of non-linearities in the data is explored, making use of a radial basis function (RBF) kernel, for which an alternative solution for the estimation of the kernel parameter is presented for cases with high-dimensional data. Additionally, the use of a backwards greedy approach based on the least-squares utility metric for the subset selection stage is proposed. The combination of these new ingredients results in the utility metric for unsupervised feature selection U2FS algorithm. The proposed U2FS algorithm succeeds in selecting the correct features in a simulation environment. In addition, the performance of the method on benchmark datasets is comparable to the state-of-the-art, while requiring less computational time. Moreover, unlike the state-of-the-art, U2FS does not require any tuning of parameters.

Respiratory sinus arrhythmia (RSA) is a form of cardiorespiratory coupling. It is observed as changes in the heart rate in synchrony with the respiration. RSA has been hypothesized to be due to a combination of linear and nonlinear effects. The quantification of the latter, in turn, has been suggested as a biomarker to improve the assessment of several conditions and diseases. In this study, a framework to quantify RSA using support vector machines is presented. The methods are based on multivariate autoregressive models, in which the present samples of the heart rate variability are predicted as combinations of past samples of the respiration. The selection and tuning of a kernel in these models allows to solve the regression problem taking into account only the linear components, or both the linear and the nonlinear ones. The methods are tested in simulated data as well as in a dataset of polysomnographic studies taken from 110 obstructive sleep apnea patients. In the simulation, the methods were able to capture the nonlinear components when a weak cardiorespiratory coupling occurs. When the coupling increases, the nonlinear part of the coupling is not detected and the interaction is found to be of linear nature. The trends observed in the application in real data show that, in the studied dataset, the proposed methods captured a more prominent linear interaction than the nonlinear one.

Objective: Automated seizure detection is a key aspect of wearable seizure warning systems. As a result, the quality of life of refractory epilepsy patients could be improved. Most state-of-the-art algorithms for heart rate-based seizure detection use a so-called patient-independent approach, which do not take into account patient-specific data during algorithm training. Although such systems are easy to use in practice, they lead to many false detections as the ictal heart rate changes are patient-dependent. In practice, only a limited amount of accurately annotated patient data is typically available, which makes it difficult to create fully patient-specific algorithms.Methods: In this context, this study proposes for the first time a new transfer learning approach that allows to personalize heart rate-based seizure detection by using only a couple of days of data per patient. The algorithm was evaluated on 2,172 h of single-lead ECG data from 24 temporal lobe epilepsy patients including 227 focal impaired awareness seizures.Results: The proposed personalized approach resulted in an overall sensitivity of 71% with 1.9 false detections per hour. This is an average decrease in false detection rate of 37% compared to the reference patient-independent algorithm using only a limited amount of personal seizure data. The proposed transfer learning approach adapts faster and more robustly to patient-specific characteristics than other alternatives for personalization in the literature.Conclusion: The proposed method allows an easy implementable solution to personalize heart rate-based seizure detection, which can improve the quality of life of refractory epilepsy patients when used as part of a multimodal seizure detection system.

Obstructive sleep apnea (OSA) patients would strongly benefit from comfortable home diagnosis, during which detection of wakefulness is essential. Therefore, capacitively-coupled electrocardiogram (ccECG) and bioimpedance (ccBioZ) sensors were used to record the sleep of suspected OSA patients, in parallel with polysomnography (PSG). The three objectives were quality assessment of the unobtrusive signals during sleep, prediction of sleep–wake using ccECG and ccBioZ, and detection of high-risk OSA patients. First, signal quality indicators (SQIs) determined the data coverage of ccECG and ccBioZ. Then, a multimodal convolutional neural network (CNN) for sleep–wake prediction was tested on these preprocessed ccECG and ccBioZ data. Finally, two indices derived from this prediction detected patients at risk. The data included 187 PSG recordings of suspected OSA patients, 36 (dataset “Test”) of which were recorded simultaneously with PSG, ccECG, and ccBioZ. As a result, two improvements were made compared to prior studies. First, the ccBioZ signal coverage increased significantly due to adaptation of the acquisition system. Secondly, the utility of the sleep–wake classifier increased as it became a unimodal network only requiring respiratory input. This was achieved by using data augmentation during training. Sleep–wake prediction on “Test” using PSG respiration resulted in a Cohen’s kappa (κ) of 0.39 and using ccBioZ in κ = 0.23. The OSA risk model identified severe OSA patients with a κ of 0.61 for PSG respiration and κ of 0.39 using ccBioZ (accuracy of 80.6% and 69.4%, respectively). This study is one of the first to perform sleep–wake staging on capacitively-coupled respiratory signals in suspected OSA patients and to detect high risk OSA patients based on ccBioZ. The technology and the proposed framework could be applied in multi-night follow-up of OSA patients.

Transfer entropy (TE) has been used to identify and quantify interactions between physiological systems. Different methods exist to estimate TE, but there is no consensus about which one performs best in specific applications. In this study, five methods (linear, k-nearest neighbors, fixed-binning with ranking, kernel density estimation and adaptive partitioning) were compared. The comparison was made on three simulation models (linear, nonlinear and linear + nonlinear dynamics). From the simulations, it was found that the best method to quantify the different interactions was adaptive partitioning. This method was then applied on data from a polysomnography study, specifically on the ECG and the respiratory signals (nasal airflow and respiratory effort around the thorax). The hypothesis that the linear and nonlinear components of cardio-respiratory interactions during light and deep sleep change with the sleep stage, was tested. Significant differences, after performing surrogate analysis, indicate an increased TE during deep sleep. However, these differences were found to be dependent on the type of respiratory signal and sampling frequency. These results highlight the importance of selecting the appropriate signals, estimation method and surrogate analysis for the study of linear and nonlinear cardio-respiratory interactions.

Many of the existing electrocardiogram (ECG) toolboxes focus on the derivation of heart rate variability features from RR-intervals. By doing so, they assume correct detection of the QRS-complexes. However, it is highly likely that not all detections are correct. Therefore, it is recommended to visualize the actual R-peak positions in the ECG signal and allow manual adaptations. In this paper we present R-DECO, an easy-to-use graphical user interface (GUI) for the detection and correction of R-peaks. Within R-DECO, the R-peaks are detected by using a detection algorithm which uses an envelope-based procedure. This procedure flattens the ECG and enhances the QRS-complexes. The algorithm obtained an overall sensitivity of 99.60% and positive predictive value of 99.69% on the MIT/BIH arrhythmia database. Additionally, R-DECO includes support for several input data formats for ECG signals, three basic filters, the possibility to load other R-peak locations and intuitive methods to correct ectopic, wrong, or missed heartbeats. All functionalities can be accessed via the GUI and the analysis results can be exported as Matlab or Excel files. The software is publicly available. Through its easy-to-use GUI, R-DECO allows both clinicians and researchers to use all functionalities, without previous knowledge.

The high prevalence of sleep apnea syndrome (SAS) and its direct relationship with an augmented risk of cardiovascular disease (CVD) have raised SAS as a primary public health problem. For this reason, extensive research aiming to understand the interaction between both conditions has been conducted. The advances in non-invasive autonomic nervous system (ANS) monitoring through heart rate variability (HRV) analysis have revealed an increased sympathetic dominance in subjects suffering from SAS when compared with controls. Similarly, HRV analysis of subjects with CVD suggests altered autonomic activity. In this work, we investigated the altered autonomic control in subjects suffering from SAS and CVD simultaneously when compared with SAS patients, as well as the possibility that ANS assessment may be useful for the early stage identification of cardiovascular risk in subjects with SAS. The analysis was performed over 199 subjects from two independent datasets during night-time, and the effects of the physiological response following an apneic episode, sleep stages, and respiration on HRV were taken into account. Results, as measured by HRV, suggest a decreased sympathetic dominance in those subjects suffering from both conditions, as well as in subjects with SAS that will develop CVDs, which was reflected in a significantly reduced sympathovagal balance (p < 0.05). In this way, ANS monitoring could contribute to improve screening and diagnosis, and eventually aid in the phenotyping of patients, as an altered response might have direct implications on cardiovascular health.

Kernel regression models have been used as non-parametric methods for fitting experimental data. However, due to their non-parametric nature, they belong to the so-called "black box" models, indicating that the relation between the input variables and the output, depending on the kernel selection, is unknown. In this paper we propose a new methodology to retrieve the relation between each input regressor variable and the output in a least squares support vector machine (LS-SVM) regression model. The method is based on oblique subspace projectors (ObSP), which allows to decouple the influence of input regressors on the output by including the undesired variables in the null space of the projection matrix. Such functional relations are represented by the nonlinear transformation of the input regressors, and their subspaces are estimated using appropriate kernel evaluations. We exploit the properties of ObSP in order to decompose the output of the obtained regression model as a sum of the partial nonlinear contributions and interaction effects of the input variables, we called this methodology Nonlinear ObSP (NObSP). We compare the performance of the proposed algorithm with the component selection and smooth operator (COSSO) for smoothing spline ANOVA models. We use as benchmark 2 toy examples and a real life regression model using the concrete strength dataset from the UCI machine learning repository. We showed that NObSP is able to outperform COSSO, producing stable estimations of the functional relations between the input regressors and the output, without the use of prior-knowledge. This methodology can be used in order to understand the functional relations between the inputs and the output in a regression model, retrieving the physical interpretation of the regression models.

Early detection of exposure to a toxic chemical, e.g., in a military context, can be life-saving. We propose to use machine learning techniques and multiple continuously measured physiological signals to detect exposure, and to identify the chemical agent. Such detection and identification could be used to alert individuals to take appropriate medical counter measures in time. As a first step, we evaluated whether exposure to an opioid (fentanyl) or a nerve agent (VX) could be detected in freely moving guinea pigs using features from respiration, electrocardiography (ECG) and electroencephalography (EEG), where machine learning models were trained and tested on different sets (across subject classification). Results showed this to be possible with close to perfect accuracy, where respiratory features were most relevant. Exposure detection accuracy rose steeply to over 95% correct during the first five minutes after exposure. Additional models were trained to correctly classify an exposed state as being induced either by fentanyl or VX. This was possible with an accuracy of almost 95%, where EEG features proved to be most relevant. Exposure detection models that were trained on subsets of animals generalized to subsets of animals that were exposed to other dosages of different chemicals. While future work is required to validate the principle in other species and to assess the robustness of the approach under different, realistic circumstances, our results indicate that utilizing different continuously measured physiological signals for early detection and identification of toxic agents is promising.

Impedance pneumography has been suggested as an ambulatory technique for the monitoring of respiratory diseases. However, its ambulatory nature makes the recordings more prone to noise sources. It is important that such noisy segments are identified and removed, since they could have a huge impact on the performance of data-driven decision support tools. In this study, we investigated the added value of machine learning algorithms to separate clean from noisy bio-impedance signals. We compared three approaches: a heuristic algorithm, a feature-based classification model (SVM) and a convolutional neural network (CNN). The dataset consists of 47 chronic obstructive pulmonary disease patients who performed an inspiratory threshold loading protocol. During this protocol, their respiration was recorded with a bio-impedance device and a spirometer, which served as a gold standard. Four annotators scored the signals for the presence of artefacts, based on the reference signal. We have shown that the accuracy of both machine learning approaches (SVM: 87.77 ± 2.64% and CNN: 87.20 ± 2.78%) is significantly higher, compared to the heuristic approach (84.69 ± 2.32%). Moreover, no significant differences could be observed between the two machine learning approaches. The feature-based and neural network model obtained a respective AUC of 92.77±2.95% and 92.51±1.74%. These findings show that a data-driven approach could be beneficial for the task of artefact detection in respiratory thoracic bio-impedance signals.

The electrocardiogram (ECG) is an important diagnostic tool for identifying cardiac problems. Nowadays, new ways to record ECG signals outside of the hospital are being investigated. A promising technique is capacitively coupled ECG (ccECG), which allows ECG signals to be recorded through insulating materials. However, as the ECG is no longer recorded in a controlled environment, this inevitably implies the presence of more artefacts. Artefact detection algorithms are used to detect and remove these. Typically, the training of a new algorithm requires a lot of ground truth data, which is costly to obtain. As many labelled contact ECG datasets exist, we could avoid the use of labelling new ccECG signals by making use of previous knowledge. Transfer learning can be used for this purpose. Here, we applied transfer learning to optimise the performance of an artefact detection model, trained on contact ECG, towards ccECG. We used ECG recordings from three different datasets, recorded with three recording devices. We showed that the accuracy of a contact-ECG classifier improved between 5 and 8% by means of transfer learning when tested on a ccECG dataset. Furthermore, we showed that only 20 segments of the ccECG dataset are sufficient to significantly increase the accuracy.

Cardiovascular diseases (CVD) are often characterized by their multifactorial complexity. This makes remote monitoring and ambulatory cardiac rehabilitation (CR) therapy challenging. Current wearable multimodal devices enable remote monitoring. Machine learning (ML) and artificial intelligence (AI) can help in tackling multifaceted datasets. However, for clinical acceptance, easy interpretability of the AI models is crucial. The goal of the present study was to investigate whether a multi-parameter sensor could be used during a standardized activity test to interpret functional capacity in the longitudinal follow-up of CR patients. A total of 129 patients were followed for 3 months during CR using 6-min walking tests (6MWT) equipped with a wearable ECG and accelerometer device. Functional capacity was assessed based on 6MWT distance (6MWD). Linear and nonlinear interpretable models were explored to predict 6MWD. The t-distributed stochastic neighboring embedding (t-SNE) technique was exploited to embed and visualize high dimensional data. The performance of support vector machine (SVM) models, combining different features and using different kernel types, to predict functional capacity was evaluated. The SVM model, using chronotropic response and effort as input features, showed a mean absolute error of 42.8 m (±36.8 m). The 3D-maps derived using the t-SNE technique visualized the relationship between sensor-derived biomarkers and functional capacity, which enables tracking of the evolution of patients throughout the CR program. The current study showed that wearable monitoring combined with interpretable ML can objectively track clinical progression in a CR population. These results pave the road towards ambulatory CR.

There exists a technological momentum towards the development of unobtrusive, simple, and reliable systems for long-term sleep monitoring. An off-the-shelf commercial pressure sensor meeting these requirements is the Emfit QS. First, the potential for sleep apnea screening was investigated by revealing clusters of contaminated and clean segments. A relationship between the irregularity of the data and the sleep apnea severity class was observed, which was valuable for screening (sensitivity 0.72, specificity 0.70), although the linear relation was limited ( R 2 of 0.16). Secondly, the study explored the suitability of this commercial sensor to be merged with gold standard polysomnography data for future sleep monitoring. As polysomnography (PSG) and Emfit signals originate from different types of sensor modalities, they cannot be regarded as strictly coupled. Therefore, an automated synchronization procedure based on artefact patterns was developed. Additionally, the optimal position of the Emfit for capturing respiratory and cardiac information similar to the PSG was identified, resulting in a position as close as possible to the thorax. The proposed approach demonstrated the potential for unobtrusive screening of sleep apnea patients at home. Furthermore, the synchronization framework enabled supervised analysis of the commercial Emfit sensor for future sleep monitoring, which can be extended to other multi-modal systems that record movements during sleep.

Clinical data is comprised by a large number of synchronously collected biomedical signals that are measured at different locations. Deciphering the interrelationships of these signals can yield important information about their dependence providing some useful clinical diagnostic data. For instance, by computing the coupling between Near-Infrared Spectroscopy signals (NIRS) and systemic variables the status of the hemodynamic regulation mechanisms can be assessed. In this paper we introduce an algorithm for the decomposition of NIRS signals into additive components. The algorithm, SIgnal DEcomposition base on Obliques Subspace Projections (SIDE-ObSP), assumes that the measured NIRS signal is a linear combination of the systemic measurements, following the linear regression model y = Ax + _. SIDE-ObSP decomposes the output such that, each component in the decomposition represents the sole linear influence of one corresponding regressor variable. This decomposition scheme aims at providing a better understanding of the relation between NIRS and systemic variables, and to provide a framework for the clinical interpretation of regression algorithms, thereby, facilitating their introduction into clinical practice. SIDE-ObSP combines oblique subspace projections (ObSP) with the structure of a mean average system in order to define adequate signal subspaces. To guarantee smoothness in the estimated regression parameters, as observed in normal physiological processes, we impose a Tikhonov regularization using a matrix differential operator. We evaluate the performance of SIDE-ObSP by using a synthetic dataset, and present two case studies in the field of cerebral hemodynamics monitoring using NIRS. In addition, we compare the performance of this method with other system identification techniques. In the first case study data from 20 neonates during the first three days of life was used, here SIDE-ObSP decoupled the influence of changes in arterial oxygen saturation from the NIRS measurements, facilitating the use of NIRS as a surrogate measure for cerebral blood flow. The second case study used data from a 3-years old infant under Extra Corporeal Membrane Oxygenation (ECMO), here SIDE-ObSP composed cerebral/peripheral tissue oxygenation, as a sum of the partial contributions from different systemic variables, facilitating the comparison between the effects of each systemic variable on the cerebral/peripheral hemodynamics.

Sleep-related conditions require high-cost and low-comfort diagnosis at the hospital during one night or longer. To overcome this situation, this work aims to evaluate an unobtrusive monitoring technique for sleep apnea. This paper presents, for the first time, the evaluation of contactless capacitively-coupled electrocardiography (ccECG) signals for the extraction of sleep apnea features, together with a comparison of different signal quality indicators. A multichannel ccECG system is used to collect signals from 15 subjects in a sleep environment from different positions. Reference quality labels were assigned for every 30-s segment. Quality indicators were calculated, and their signal classification performance was evaluated. Features for the detection of sleep apnea were extracted from capacitive and reference signals. Sleep apnea features related to heart rate and heart rate variability achieved high similarity to the reference values, with p-values of 0.94 and 0.98, which is in line with the more than 95% beat-matching obtained. Features related to signal morphology presented lower similarity with the reference, although signal similarity metrics of correlation and coherence were relatively high. Quality-based automatic classification of the signals had a maximum accuracy of 91%. Best-performing quality indicators were based on template correlation and beat-detection. Results suggest that using unobtrusive cardiac signals for the automatic detection of sleep apnea can achieve similar performance as contact signals, and indicates clinical value of ccECG. Moreover, signal segments can automatically be classified by the proposed quality metrics as a pre-processing step. Including contactless respiration signals is likely to improve the performance and provide a complete unobtrusive cardiorespiratory monitoring solution; this is a promising alternative that will allow the screening of more patients with higher comfort, for a longer time, and at a reduced cost.