Objective gait assessment in Parkinson’s Disease (PD) is limited by the absence of large, diverse, and clinically annotated motion datasets. We introduce Care-PD, the largest publicly available archive of 3D mesh gait data for PD, and the first multi-site collection spanning 9 cohorts from 8 clinical centers. All recordings (RGB video or motion capture) are converted into anonymized SMPL meshes via a harmonized preprocessing pipeline. Care-PD supports two key benchmarks: supervised clinical score prediction (estimating Unified Parkinson’s Disease Rating Scale, UPDRS, gait scores) and unsupervised motion pretext tasks (2D-to-3D keypoint lifting and full-body 3D reconstruction). Clinical prediction is evaluated under four generalization protocols: within-dataset, cross-dataset, leave-one-dataset-out, and multi-dataset in-domain adaptation.To assess clinical relevance, we compare state-of-the-art motion encoders with a traditional gait-feature baseline, finding that encoders consistently outperform handcrafted features. Pretraining on Care-PD reduces MPJPE (from 60.8mm to 7.5mm) and boosts PD severity macro-F1 by 17\%, underscoring the value of clinically curated, diverse training data. Care-PD and all benchmark code are released for non-commercial research (Code, Data).

Atrial fibrillation (AF) is a common cardiac arrhythmia with serious health consequences if not detected and treated early. Detecting AF using wearable devices with photoplethysmography (PPG) sensors and deep neural networks has demonstrated some success using proprietary algorithms in commercial solutions. However, to improve continuous AF detection in ambulatory settings towards a population-wide screening use case, we face several challenges, one of which is the lack of large-scale labeled training data. To address this challenge, we propose to leverage AF alarms from bedside patient monitors to label concurrent PPG signals, resulting in the largest PPG-AF dataset so far (8.5 M 30-second records from 24,100 patients) and demonstrating a practical approach to build large labeled PPG datasets. Furthermore, we recognize that the AF labels thus obtained contain errors because of false AF alarms generated from imperfect built-in algorithms from bedside monitors. Dealing with label noise with unknown distribution characteristics in this case requires advanced algorithms. We, therefore, introduce and open-source a novel loss design, the cluster membership consistency (CMC) loss, to mitigate label errors. By comparing CMC with state-of-the-art methods selected from a noisy label competition, we demonstrate its superiority in handling label noise in PPG data, resilience to poor-quality signals, and computational efficiency.

OBJECTIVE: Psychiatric evaluation suffers from subjectivity and bias, and is hard to scale due to intensive professional training requirements. In this work, we investigated whether behavioral and physiological signals, extracted from tele-video interviews, differ in individuals with psychiatric disorders. METHODS: Temporal variations in facial expression, vocal expression, linguistic expression, and cardiovascular modulation were extracted from simultaneously recorded audio and video of remote interviews. Averages, standard deviations, and Markovian process-derived statistics of these features were computed from 73 subjects. Four binary classification tasks were defined: detecting 1) any clinically-diagnosed psychiatric disorder, 2) major depressive disorder, 3) self-rated depression, and 4) self-rated anxiety. Each modality was evaluated individually and in combination. RESULTS: Statistically significant feature differences were found between psychiatric and control subjects. Correlations were found between features and self-rated depression and anxiety scores. Heart rate dynamics provided the best unimodal performance with areas under the receiver-operator curve (AUROCs) of 0.68-0.75 (depending on the classification task). Combining multiple modalities provided AUROCs of 0.72-0.82. CONCLUSION: Multimodal features extracted from remote interviews revealed informative characteristics of clinically diagnosed and self-rated mental health status. SIGNIFICANCE: The proposed multimodal approach has the potential to facilitate scalable, remote, and low-cost assessment for low-burden automated mental health services.

Poor sleep quality in Autism Spectrum Disorder (ASD) individuals is linked to severe daytime behaviors. This study explores the relationship between a prior night's sleep structure and its predictive power for next-day behavior in ASD individuals. The motion was extracted using a low-cost near-infrared camera in a privacy-preserving way. Over two years, we recorded overnight data from 14 individuals, spanning over 2000 nights, and tracked challenging daytime behaviors, including aggression, self-injury, and disruption. We developed an ensemble machine learning algorithm to predict next-day behavior in the morning and the afternoon. Our findings indicate that sleep quality is a more reliable predictor of morning behavior than afternoon behavior the next day. The proposed model attained an accuracy of 74% and a F1 score of 0.74in target-sensitive tasks and 67% accuracy and 0.69 F1 score in target-insensitive tasks. For 7 of the 14, better-than-chance balanced accuracy was obtained (p-value 0.05), with 3 showing significant trends (p-value 0.1). These results suggest off-body, privacy-preserving sleep monitoring as a viable method for predicting next-day adverse behavior in ASD individuals, with the potential for behavioral intervention and enhanced care in social and learning settings.

False arrhythmia alarms in intensive care units (ICUs) are a continuing problem despite considerable effort from industrial and academic algorithm developers. Of all life-threatening arrhythmias, ventricular tachycardia (VT) stands out as the most challenging arrhythmia to detect reliably. We introduce a new annotated VT alarm database, VTaC (Ventricular Tachycardia annotated alarms from ICUs) consisting of over 5,000 waveform recordings with VT alarms triggered by bedside monitors in the ICU. Each VT alarm waveform in the dataset has been labeled by at least two independent human expert annotators. The dataset encompasses data collected from ICUs in two major US hospitals and includes data from three leading bedside monitor manufacturers, providing a diverse and representative collection of alarm waveform data. Each waveform recording comprises at least two electrocardiogram (ECG) leads and one or more pulsatile waveforms, such as photoplethysmogram (PPG or PLETH) and arterial blood pressure (ABP) waveforms. We demonstrate the utility of this new benchmark dataset for the task of false arrhythmia alarm reduction, and present performance of multiple machine learning approaches, including conventional supervised machine learning, deep learning, semi-supervised learning, and generative approaches for the task of VT false alarm reduction.

OBJECTIVE: Murmurs are abnormal heart sounds, identified by experts through cardiac auscultation. The murmur grade, a quantitative measure of the murmur intensity, is strongly correlated with the patient's clinical condition. This work aims to estimate each patient's murmur grade (i.e., absent, soft, loud) from multiple auscultation location phonocardiograms (PCGs) of a large population of pediatric patients from a low-resource rural area. METHODS: The Mel spectrogram representation of each PCG recording is given to an ensemble of 15 convolutional residual neural networks with channel-wise attention mechanisms to classify each PCG recording. The final murmur grade for each patient is derived based on the proposed decision rule and considering all estimated labels for available recordings. The proposed method is cross-validated on a dataset consisting of 3456 PCG recordings from 1007 patients using a stratified ten-fold cross-validation. Additionally, the method was tested on a hidden test set comprised of 1538 PCG recordings from 442 patients. RESULTS: The overall cross-validation performances for patient-level murmur gradings are 86.3% and 81.6% in terms of the unweighted average of sensitivities and F1-scores, respectively. The sensitivities (and F1-scores) for absent, soft, and loud murmurs are 90.7% (93.6%), 75.8% (66.8%), and 92.3% (84.2%), respectively. On the test set, the algorithm achieves an unweighted average of sensitivities of 80.4% and an F1-score of 75.8%. CONCLUSIONS: This study provides a potential approach for algorithmic pre-screening in low-resource settings with relatively high expert screening costs. SIGNIFICANCE: The proposed method represents a significant step beyond detection of murmurs, providing characterization of intensity, which may provide an enhanced classification of clinical outcomes.

Assessing fetal development is essential to the provision of healthcare for both mothers and fetuses. In low- and middle-income countries, conditions that increase the risk of fetal growth restriction (FGR) are often more prevalent. In these regions, barriers to accessing healthcare and social services exacerbate fetal maternal health problems. One of these barriers is the lack of affordable diagnostic technologies. To address this issue, this work introduces an end-to-end algorithm applied to a low-cost, hand-held Doppler ultrasound device for estimating gestational age (GA), and by inference, FGR. The Doppler ultrasound signals used in this study were collected from 226 pregnancies (45 low birth weight at delivery) between 5 and 9 months GA by lay midwives in highland Guatemala. We designed a hierarchical deep sequence learning model with an attention mechanism to learn the normative dynamics of fetal cardiac activity in different stages of development. This resulted in a state-of-the-art GA estimation performance, with an average error of 0.79 months. This is close to the theoretical minimum for the given quantization level of one month. The model was then tested on Doppler recordings of the fetuses with low birth weight and the estimated GA was shown to be lower than the GA calculated from last menstruation. Thus, this could be interpreted as a potential sign of developmental retardation (or FGR) associated with low birth weight, and referral and intervention may be necessary.

Cardiac auscultation is one of the most cost-effective techniques used to detect and identify many heart conditions. Computer-assisted decision systems based on auscultation can support physicians in their decisions. Unfortunately, the application of such systems in clinical trials is still minimal since most of them only aim to detect the presence of extra or abnormal waves in the phonocardiogram signal, i.e., only a binary ground truth variable (normal vs abnormal) is provided. This is mainly due to the lack of large publicly available datasets, where a more detailed description of such abnormal waves (e.g., cardiac murmurs) exists. To pave the way to more effective research on healthcare recommendation systems based on auscultation, our team has prepared the currently largest pediatric heart sound dataset. A total of 5282 recordings have been collected from the four main auscultation locations of 1568 patients, in the process, 215780 heart sounds have been manually annotated. Furthermore, and for the first time, each cardiac murmur has been manually annotated by an expert annotator according to its timing, shape, pitch, grading, and quality. In addition, the auscultation locations where the murmur is present were identified as well as the auscultation location where the murmur is detected more intensively. Such detailed description for a relatively large number of heart sounds may pave the way for new machine learning algorithms with a real-world application for the detection and analysis of murmur waves for diagnostic purposes.

Epilepsy is the second most common neurological disorder, affecting 65 million people around the world. It is associated with seizures - a sudden, uncontrolled electrical disturbance in the brain that can lead to profound transient changes in behavior, movements, feelings, and levels of consciousness. Current approaches to developing a generalized automated seizure detection algorithm rely on constructing large, labeled training and test corpora of electroencephalograms (EEGs) from different individuals. However, due to the inherent inter-subject variability, heterogeneity of acquisition hardware, different montage choices, and various recording environments, EEG patterns may exhibit very different distributions over time and between individuals. Therefore, training an algorithm on such data without accounting for this diversity can affect the performance of any classifier or predictor. In addition, this process smooths out individual differences, producing a general, but a non-specific model. To address these issues, we propose a novel method that generates transferable features to interpolate between features from the training and test sets. This is achieved by adversarially training deep classifiers to make consistent classifications or predictions over the transferable features. Experiments on an EEG seizure databases demonstrate that the proposed method increases the accuracy over state-of-the-art from 86.83% to 91.71% and specificity from 87.38% to 94.73% while reducing the false positive rate/hour from 0.8/hour to 0.58/hour. Therefore, this work has the potential for significantly reducing workload in reviewing clinical EEGs for seizures, and for improved real-time closed-loop vagal stimulation.

Post-Traumatic Stress Disorder (PTSD) is a psychiatric condition resulting from threatening or horrifying events. We hypothesized that circadian rhythm changes, measured by a wrist-worn research watch are predictive of post-trauma outcomes. APPROACH: 1618 post-trauma patients were enrolled after admission to emergency departments (ED). Three standardized questionnaires were administered at week eight to measure post-trauma outcomes related to PTSD, sleep disturbance, and pain interference with daily life. Pulse activity and movement data were captured from a research watch for eight weeks. Standard and novel movement and cardiovascular metrics that reflect circadian rhythms were derived using this data. These features were used to train different classifiers to predict the three outcomes derived from week-eight surveys. Clinical surveys administered at ED were also used as features in the baseline models. RESULTS: The highest cross-validated performance of research watch-based features was achieved for classifying participants with pain interference by a logistic regression model, with an area under the receiver operating characteristic curve (AUC) of 0.70. The ED survey-based model achieved an AUC of 0.77, and the fusion of research watch and ED survey metrics improved the AUC to 0.79. SIGNIFICANCE: This work represents the first attempt to predict and classify post-trauma symptoms from passive wearable data using machine learning approaches that leverage the circadian desynchrony in a potential PTSD population.

Psychometric measures reflecting people’s knowledge, ability, attitudes, and personality traits are critical for many real-world applications, such as e-commerce, health care, and cybersecurity. However, traditional methods cannot collect and measure rich psychometric dimensions in a timely and unobtrusive manner. Consequently, despite their importance, psychometric dimensions have received limited attention from the natural language processing and information retrieval communities. In this article, we propose a deep learning architecture, PyNDA, to extract psychometric dimensions from user-generated texts. PyNDA contains a novel representation embedding, a demographic embedding, a structural equation model (SEM) encoder, and a multitask learning mechanism designed to work in unison to address the unique challenges associated with extracting rich, sophisticated, and user-centric psychometric dimensions. Our experiments on three real-world datasets encompassing 11 psychometric dimensions, including trust, anxiety, and literacy, show that PyNDA markedly outperforms traditional feature-based classifiers as well as the state-of-the-art deep learning architectures. Ablation analysis reveals that each component of PyNDA significantly contributes to its overall performance. Collectively, the results demonstrate the efficacy of the proposed architecture for facilitating rich psychometric analysis. Our results have important implications for user-centric information extraction and retrieval systems looking to measure and incorporate psychometric dimensions.

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.

Heart failure (HF) is one of the leading causes of mortality in the United States with a high economic burden due to readmissions. We present a novel approach to remotely monitor quality of life in patients with HF using a smartphone app and a scalable cloud-based architecture. In a preliminary study, we assess continuous data from 10 HF subjects over a period of up to a year. Over 680 million samples of physical movement data, 9,000 geographic location updates, and 11,000 individual social networking events in the form of phone calls were captured from the app. Personalized models were constructed from these data to estimate self-reported quality of life using the Kansas City Cardiomyopathy Questionnaire (KCCQ), which has been shown to be a reliable health status measure for HF patients. Generalized linear models using only activity features were shown to reliably estimate the KCCQ score with an out of sample mean absolute error of 5.71%. Personalized models for estimating the HF severity as mild or severe were also built as a proof of concept to detect when a subject's data indicated a clinical deterioration. Average out of sample accuracy was 83% for this binary classification problem. Creation of personalized models from passive smartphone data collected `in-the-wild' to identify changes in HF severity appears possible. This new approach holds promise as a low burden and accurate method of monitoring HF symptoms, which could aid clinicians in early assessment and prevention of adverse outcomes.

The scalability of medical technology in low resource settings requires a higher level of usability and clear decision support compared to conventional devices, since users often have very limited training. In particular, it is important to provide users with real time feedback on data quality during the patient information acquisition in a manner that enables the user to take immediate corrective action.

Detection of atrial fibrillation (AF), a type of cardiac arrhythmia, is difficult since many cases of AF are usually clinically silent and undiagnosed. In particular paroxysmal AF is a form of AF that occurs occasionally, and has a higher probability of being undetected. In this work, we present an attention based deep learning framework for detection of paroxysmal AF episodes from a sequence of windows. Time-frequency representation of 30 seconds recording windows, over a 10 minute data segment, are fed sequentially into a deep convolutional neural network for image-based feature extraction, which are then presented to a bidirectional recurrent neural network with an attention layer for AF detection. To demonstrate the effectiveness of the proposed framework for transient AF detection, we use a database of 24 hour Holter Electrocardiogram (ECG) recordings acquired from 2850 patients at the University of Virginia heart station. The algorithm achieves an AUC of 0.94 on the testing set, which exceeds the performance of baseline models. We also demonstrate the cross-domain generalizablity of the approach by adapting the learned model parameters from one recording modality (ECG) to another (photoplethysmogram) with improved AF detection performance. The proposed high accuracy, low false alarm algorithm for detecting paroxysmal AF has potential applications in long-term monitoring using wearable sensors.

Clinical data management systems typically provide caregiver teams with useful information, derived from large, sometimes highly heterogeneous, data sources that are often changing dynamically. Over the last decade there has been a significant surge in interest in using these data sources, from simply re-using the standard clinical databases for event prediction or decision support, to including dynamic and patient-specific information into clinical monitoring and prediction problems. However, in most cases, commercial clinical databases have been designed to document clinical activity for reporting, liability and billing reasons, rather than for developing new algorithms. With increasing excitement surrounding "secondary use of medical records" and "Big Data" analytics, it is important to understand the limitations of current databases and what needs to change in order to enter an era of "precision medicine." This review article covers many of the issues involved in the collection and preprocessing of critical care data. The three challenges in critical care are considered: compartmentalization, corruption, and complexity. A range of applications addressing these issues are covered, including the modernization of static acuity scoring; on-line patient tracking; personalized prediction and risk assessment; artifact detection; state estimation; and incorporation of multimodal data sources such as genomic and free text data.

Obstructive sleep apnoea (OSA) is a sleep disorder with long-term consequences. Long-term effects include sleep-related issues and cardiovascular diseases. OSA is often diagnosed with an overnight sleep test called a polysomnogram. Monitoring can be costly with long wait times for diagnosis. In this paper, a novel OSA screening framework and prototype phone application are introduced. A database of 856 patients that underwent at-home polygraphy was collected. Features were derived from audio, actigraphy, photoplethysmography (PPG), and demographics, and used as the inputs of a support vector machine (SVM) classifier. The SVM was trained on 735 patients and tested on 121 patients. Classification on the test set had an accuracy of up to 92.2% when classifying subjects as having moderate or severe OSA versus being healthy or a snorer based on the clinicians' diagnoses. The signal processing and machine learning algorithms were ported to Java and integrated into the phone application-SleepAp. SleepAp records the body position, audio, actigraphy and PPG signals, and implements the clinically validated STOP-BANG questionnaire. It derives features from the signals and classifies the user as having OSA or not using the SVM trained on the clinical database. The resulting software could provide a new, easy-to-use, low-cost, and widely available modality for OSA screening.

There is potential for the use of mobile phones to remotely identify patients with a high risk of heart conditions using automated auscultation. However, accurate heart sound analysis is dependent on the quality of heart sound recordings. This paper investigates the signal quality classification of phonocardiograms (PCGs) recorded on two devices (a 3M Littmann 3200 electronic stethoscope and an iPhone 3G). These recordings were professionally annotated and classified using a support vector machine (SVM) and a combination of ten signal quality metrics computed from each recording as input features. One third of all mobile phone-recorded PCGs were found to be of high quality. The classifier was able to distinguish good and bad-quality iPhone recordings with 87.0% accuracy, the Littmann recordings with accuracy of 76.4% and the combined set with accuracy of 85.6% on unseen test data. Therefore, the quality of PCGs made with a range of stethoscopes can be accurately classified using this technique.