Physiologically-Informed Gaussian Processes for Interpretable Modelling of Psycho-Physiological States.

The widespread popularity of Machine Learning (ML) models in healthcare solutions has increased the demand for their interpretability and accountability. In this paper, we propose the Physiologically-Informed Gaussian Process (PhGP) classification model, an interpretable machine learning model founded on the Bayesian nature of Gaussian Processes (GPs). Specifically, we inject problem-specific domain knowledge of inherent physiological mechanisms underlying the psycho-physiological states as a prior distribution over the GP latent space. Thus, to estimate the hyper-parameters in PhGP, we rely on the information from raw physiological signals as well as the designed prior function encoding the physiologically-inspired modelling assumptions. Alongside this new model, we present novel interpretability metrics that highlight the most informative input regions that contribute to the GP prediction. We evaluate the ability of PhGP to provide an accurate and interpretable classification on three different datasets, including electrodermal activity (EDA) signals collected during emotional, painful, and stressful tasks. Our results demonstrate that, for all three tasks, recognition performance is improved by using the PhGP model compared to competitive methods. Moreover, PhGP is able to provide physiological sound interpretations over its predictions.

Improving Classification of Tetanus Severity for Patients in Low-Middle Income Countries Wearing ECG Sensors by Using a CNN-Transformer Network.

Tetanus is a life-threatening infectious disease, which is still common in low- and middle-income countries, including in Vietnam. This disease is characterized by muscle spasm and in severe cases is complicated by autonomic dysfunction. Ideally continuous vital sign monitoring using bedside monitors allows the prompt detection of the onset of autonomic nervous system dysfunction or avoiding rapid deterioration. Detection can be improved using heart rate variability analysis from ECG signals. Recently, characteristic ECG and heart rate variability features have been shown to be of value in classifying tetanus severity. However, conventional manual analysis of ECG is time-consuming. The traditional convolutional neural network (CNN) has limitations in extracting the global context information, due to its fixed-sized kernel filters. In this work, we propose a novel hybrid CNN-Transformer model to automatically classify tetanus severity using tetanus monitoring from low-cost wearable sensors. This model can capture the local features from the CNN and the global features from the Transformer. The time series imaging - spectrogram - is transformed from one-dimensional ECG signal and input to the proposed model. The CNN-Transformer model outperforms state-of-the-art methods in tetanus classification, achieves results with a F1 score of 0.82±0.03, precision of 0.94±0.03, recall of 0.73±0.07, specificity of 0.97±0.02, accuracy of 0.88±0.01 and AUC of 0.85±0.03. In addition, we found that Random Forest with enough manually selected features can be comparable with the proposed CNN-Transformer model.

Classification of Tetanus Severity in Intensive-Care Settings for Low-Income Countries Using Wearable Sensing.

Infectious diseases remain a common problem in low- and middle-income countries, including in Vietnam. Tetanus is a severe infectious disease characterized by muscle spasms and complicated by autonomic nervous system dysfunction in severe cases. Patients require careful monitoring using electrocardiograms (ECGs) to detect deterioration and the onset of autonomic nervous system dysfunction as early as possible. Machine learning analysis of ECG has been shown of extra value in predicting tetanus severity, however any additional ECG signal analysis places a high demand on time-limited hospital staff and requires specialist equipment. Therefore, we present a novel approach to tetanus monitoring from low-cost wearable sensors combined with a deep-learning-based automatic severity detection. This approach can automatically triage tetanus patients and reduce the burden on hospital staff. In this study, we propose a two-dimensional (2D) convolutional neural network with a channel-wise attention mechanism for the binary classification of ECG signals. According to the Ablett classification of tetanus severity, we define grades 1 and 2 as mild tetanus and grades 3 and 4 as severe tetanus. The one-dimensional ECG time series signals are transformed into 2D spectrograms. The 2D attention-based network is designed to extract the features from the input spectrograms. Experiments demonstrate a promising performance for the proposed method in tetanus classification with an F1 score of 0.79 ± 0.03, precision of 0.78 ± 0.08, recall of 0.82 ± 0.05, specificity of 0.85 ± 0.08, accuracy of 0.84 ± 0.04 and AUC of 0.84 ± 0.03.

Sepsis Mortality Prediction Using Wearable Monitoring in Low-Middle Income Countries.

Sepsis is associated with high mortality-particularly in low-middle income countries (LMICs). Critical care management of sepsis is challenging in LMICs due to the lack of care providers and the high cost of bedside monitors. Recent advances in wearable sensor technology and machine learning (ML) models in healthcare promise to deliver new ways of digital monitoring integrated with automated decision systems to reduce the mortality risk in sepsis. In this study, firstly, we aim to assess the feasibility of using wearable sensors instead of traditional bedside monitors in the sepsis care management of hospital admitted patients, and secondly, to introduce automated prediction models for the mortality prediction of sepsis patients. To this end, we continuously monitored 50 sepsis patients for nearly 24 h after their admission to the Hospital for Tropical Diseases in Vietnam. We then compared the performance and interpretability of state-of-the-art ML models for the task of mortality prediction of sepsis using the heart rate variability (HRV) signal from wearable sensors and vital signs from bedside monitors. Our results show that all ML models trained on wearable data outperformed ML models trained on data gathered from the bedside monitors for the task of mortality prediction with the highest performance (area under the precision recall curve = 0.83) achieved using time-varying features of HRV and recurrent neural networks. Our results demonstrate that the integration of automated ML prediction models with wearable technology is well suited for helping clinicians who manage sepsis patients in LMICs to reduce the mortality risk of sepsis.

Quantifying multidimensional control mechanisms of cardiovascular dynamics during multiple concurrent stressors.

Heartbeat regulation is achieved through different routes originating from central autonomic network sources, as well as peripheral control mechanisms. While previous studies successfully characterized cardiovascular regulatory mechanisms during a single stressor, to the best of our knowledge, a combination of multiple concurrent elicitations leading to the activation of different autonomic regulatory routes has not been investigated yet. Therefore, in this study, we propose a novel modeling framework for the quantification of heartbeat regulatory mechanisms driven by different neural routes. The framework is evaluated using two heartbeat datasets gathered from healthy subjects undergoing physical and mental stressors, as well as their concurrent administration. Experimental results indicate that more than 70% of the heartbeat regulatory dynamics is driven by the physical stressor when combining physical and cognitive/emotional stressors. The proposed framework provides quantitative insights and novel perspectives for neural activity on cardiac control dynamics, likely highlighting new biomarkers in the psychophysiology and physiopathology fields. A Matlab implementation of the proposed tool is available online.

Increased functional connectivity within alpha and theta frequency bands in dysphoria: A resting-state EEG study.

BACKGROUND: The understanding of neurophysiological correlates underlying the risk of developing depression may have a significant impact on its early and objective identification. Research has identified abnormal resting-state electroencephalography (EEG) power and functional connectivity patterns in major depression. However, the entity of dysfunctional EEG dynamics in dysphoria is yet unknown. METHODS: 32-channel EEG was recorded in 26 female individuals with dysphoria and in 38 age-matched, female healthy controls. EEG power spectra and alpha asymmetry in frontal and posterior channels were calculated in a 4-minute resting condition. An EEG functional connectivity analysis was conducted through phase locking values, particularly mean phase coherence. RESULTS: While individuals with dysphoria did not differ from controls in EEG spectra and asymmetry, they exhibited dysfunctional brain connectivity. Particularly, in the theta band (4-8 Hz), participants with dysphoria showed increased connectivity between right frontal and central areas and right temporal and left occipital areas. Moreover, in the alpha band (8-12 Hz), dysphoria was associated with increased connectivity between right and left prefrontal cortex and between frontal and central-occipital areas bilaterally. LIMITATIONS: All participants belonged to the female gender and were relatively young. Mean phase coherence did not allow to compute the causal and directional relation between brain areas. CONCLUSIONS: An increased EEG functional connectivity in the theta and alpha bands characterizes dysphoria. These patterns may be associated with the excessive self-focus and ruminative thinking that typifies depressive symptoms. EEG connectivity patterns may represent a promising measure to identify individuals with a higher risk of developing depression.

Assessing Autonomic Function from Electrodermal Activity and Heart Rate Variability During Cold-Pressor Test and Emotional Challenge.

Standard functional assessment of autonomic nervous system (ANS) activity on cardiovascular control relies on spectral analysis of heart rate variability (HRV) series. However, difficulties in obtaining a reliable measure of sympathetic activity from HRV spectra limits the exploitation of sympatho-vagal metrics. On the other hand, measures of electrodermal activity (EDA) have been demonstrated to provide a reliable quantifier of sympathetic dynamics. In this study we propose novel indices of phasic autonomic regulation mechanisms by combining HRV and EDA correlates and thoroughly investigating their time-varying dynamics. HRV and EDA series were gathered from 26 healthy subjects during a cold-pressor test and emotional stimuli. Instantaneous linear and nonlinear (bispectral) estimates of vagal dynamics were obtained from HRV through inhomogeneous point-process models, and combined with a sensitive maker of sympathetic tone from EDA spectral power. A wavelet decomposition analysis was applied to estimate phasic components of the proposed sympatho-vagal indices. Results show significant statistical differences for the proposed indices between the cold-pressor elicitation and previous resting state. Furthermore, an accuracy of 73.08% was achieved for the automatic emotional valence recognition. The proposed nonlinear processing of phasic ANS markers brings novel insights on autonomic functioning that can be exploited in the field of affective computing and psychophysiology.

The Role of Haptic Stimuli on Affective Reading: a Pilot Study.

The affective role of touch has opened new perspectives in human-machine interaction. This paper presents an emotion recognition algorithm to investigate the role of tactile stimuli conveyed through a wearable haptic system during affective reading. To this end, a group of 32 healthy volunteers underwent an emotional stimulation by reading affective texts, with and without the concurrent presence of pleasant haptic stimuli. Throughout the experiment, autonomic nervous system dynamics was quantified through heart rate variability (HRV) and electrodermal activity (EDA) analyses. EDA and HRV features were then used as input of a SVM-RFE learning algorithm for an automatic recognition of neutral and arousing texts. The affective recognition of the reading was performed in the presence or absence of the haptic stimulation. Results show that the affective perception induced by the neutral and arousing reading were discriminated with a significantly improved accuracy (+14.5%) when a caress-like haptic stimulus was conveyed to the user.

Detection of pathological heart sounds.

Heart sound analysis has been a major topic of research over the past few decades. However, the necessity for a large and reliable database has been a major concern in these studies. OBJECTIVE: Noting that the current heart sound classification methods do not work properly for noisy signals, the PhysioNet/CinC Challenge 2016 aims to develop the heart sound classification algorithms by providing a global open database for challengers. This paper addresses the problem of heart sound classification methods within noisy real-world phonocardiogram recordings by implementing an innovative approach. SIGNIFICANCE: After locating the fundamental heart sounds and the systolic and diastolic components, a novel method named cycle quality assessment is applied to each recording. The presented method detects those cycles which are less affected by noise and better segmented by the use of two criteria here proposed in this paper. The selected cycles are the inputs of a further feature extraction process. APPROACH: Due to the variability of the heart sound signal induced by various cardiac arrhythmias, four sets of features from the time, time-frequency and perceptual domains are extracted. Before starting the main classification process, the obtained 90-dimensional feature vector is mapped to a new feature space to pre-detect normal recordings by applying a Fisher's discriminant analysis. The main classification procedure is then done based on three feed-forward neural networks and a voting system among classifiers. MAIN RESULTS: The presented method is evaluated using the training and hidden test sets of the PhysioNet/CinC Challenge 2016. Also, the results are compared with the top five ranked submissions. The results indicate that the proposed method is effective in classifying heart sounds as normal versus abnormal recordings.