OBJECTIVE: To design and validate a novel deep generative model for seismocardiogram (SCG) dataset augmentation. SCG is a noninvasively acquired cardiomechanical signal used in a wide range of cardivascular monitoring tasks; however, these approaches are limited due to the scarcity of SCG data. METHODS: A deep generative model based on transformer neural networks is proposed to enable SCG dataset augmentation with control over features such as aortic opening (AO), aortic closing (AC), and participant-specific morphology. We compared the generated SCG beats to real human beats using various distribution distance metrics, notably Sliced-Wasserstein Distance (SWD). The benefits of dataset augmentation using the proposed model for other machine learning tasks were also explored. RESULTS: Experimental results showed smaller distribution distances for all metrics between the synthetically generated set of SCG and a test set of human SCG, compared to distances from an animal dataset (1.14× SWD), Gaussian noise (2.5× SWD), or other comparison sets of data. The input and output features also showed minimal error (95% limits of agreement for pre-ejection period [PEP] and left ventricular ejection time [LVET] timings are 0.03 ± 3.81 ms and -0.28 ± 6.08 ms, respectively). Experimental results for data augmentation for a PEP estimation task showed 3.3% accuracy improvement on an average for every 10% augmentation (ratio of synthetic data to real data). CONCLUSION: The model is thus able to generate physiologically diverse, realistic SCG signals with precise control over AO and AC features. This will uniquely enable dataset augmentation for SCG processing and machine learning to overcome data scarcity.

We employed wearable multimodal sensing (heart rate and triaxial accelerometry) with machine learning to enable early prediction of impending exertional heat stroke (EHS). US Army Rangers and Combat Engineers (N = 2,102) were instrumented while participating in rigorous 7-mile and 12-mile loaded rucksack timed marches. There were three EHS cases, and data from 478 Rangers were analyzed for model building and controls. The data-driven machine learning approach incorporated estimates of physiological strain (heart rate) and physical stress (estimated metabolic rate) trajectories, followed by reconstruction to obtain compressed representations which then fed into anomaly detection for EHS prediction. Impending EHS was predicted from 33 to 69 min before collapse. These findings demonstrate that low dimensional physiological stress to strain patterns with machine learning anomaly detection enables early prediction of impending EHS which will allow interventions that minimize or avoid pathophysiological sequelae. We describe how our approach can be expanded to other physical activities and enhanced with novel sensors.

Hypovolemia remains the leading cause of preventable death in trauma cases. Recent research has demonstrated that using noninvasive continuous waveforms rather than traditional vital signs improves accuracy in early detection of hypovolemia to assist in triage and resuscitation. This work evaluates random forest models trained on different subsets of data from a pig model (n = 6) of absolute (bleeding) and relative (nitroglycerin-induced vasodilation) progressive hypovolemia (to 20% decrease in mean arterial pressure) and resuscitation. Features for the models were derived from a multi-modal set of wearable sensors, comprised of the electrocardiogram (ECG), seismocardiogram (SCG) and reflective photoplethysmogram (RPPG) and were normalized to each subject.s baseline. The median RMSE between predicted and actual percent progression towards cardiovascular decompensation for the best model was 30.5% during the relative period, 16.8% during absolute and 22.1% during resuscitation. The least squares best fit line over the mean aggregated predictions had a slope of 0.65 and intercept of 12.3, with an R2value of 0.93. When transitioned to a binary classification problem to identify decompensation, this model achieved an AUROC of 0.80. This study: a) developed a global model incorporating ECG, SCG and RPPG features for estimating individual-specific decompensation from progressive relative and absolute hypovolemia and resuscitation; b) demonstrated SCG as the most important modality to predict decompensation; c) demonstrated efficacy of random forest models trained on different data subsets; and d) demonstrated adding training data from two discrete forms of hypovolemia increases prediction accuracy for the other form of hypovolemia and resuscitation.

Visualizing latent embeddings is a popular approach to explain classification models, including deep neural networks. However, existing visualization methods such as t-distributed Stochastic Neighbor Embedding (t-SNE) and Uniform Manifold Approximation Projection (UMAP) are often used as a post-processing step which is independent of the classification models. The resulting visualization can be misaligned with the classification models. In this paper, we propose ViVA, a novel method for semi-supervised Visualization via Variational Autoencoders. ViVA learns from both unlabeled and labeled data by jointly optimizing both visualization loss and classification loss. As a parameterized model using neural networks, ViVA can easily project new data to the same embedding space. Experiments show that ViVA can achieve better visualization quality as well as classification accuracy on multiple challenging datasets compared to several visualization baselines, including t-SNE and UMAP.

Although deep learning models trained on electronic health records (EHR) data have shown state-of-the-art performance in many predictive clinical tasks, the discovery of adversarial examples (i.e., input data that are engineered to cause misclassification) has exposed vulnerabilities with lab and imaging data. We specifically consider adversarial examples with longitudinal EHR data, an area that has not been previously examined because of the challenges with temporal high-dimensional and sparse features. We propose Longitudinal AdVersarial Attack (, a saliency score based adversarial example using a method that requires a minimal number of perturbations and that automatically minimizes the likelihood of detection. Features are selected and modified by jointly modeling a saliency map and attention mechanism. Experimental results with longitudinal EHR data show that an substantially reduce model performance for attention-based target models (from AUPR = 0.5 to AUPR = 0.08).

In order to reduce the possibility of neural injury from seizures and sidestep the need for a neurologist to spend hours on manually reviewing the EEG recording, it is critical to automatically detect and classify "interictal-ictal continuum" (IIC) patterns from EEG data. However, the existing IIC classification techniques are shown to be not accurate and robust enough for clinical use because of the lack of high quality labels of EEG segments as training data. Obtaining high-quality labeled data is traditionally a manual process by trained clinicians that can be tedious, time-consuming, and error-prone. In this work, we propose Smile, an industrial scale system that provides an end-to-end solution to the IIC pattern classification problem. The core components of Smile include a visualization-based time series labeling module and a deep-learning based active learning module. The labeling module enables the users to explore and label 350 million EEG segments (30TB) at interactive speed. The multiple coordinated views allow the users to examine the EEG signals from both time domain and frequency domain simultaneously. The active learning module first trains a deep neural network that automatically extracts both the local features with respect to each segment itself and the long term dynamics of the EEG signals to classify IIC patterns. Then leveraging the output of the deep learning model, the EEG segments that can best improve the model are selected and prompted to clinicians to label. This process is iterated until the clinicians and the models show high degree of agreement. Our initial experimental results show that our Smile system allows the clinicians to label the EEG segments at will with a response time below 500 ms. The accuracy of the model is progressively improved as more and more high quality labels are acquired over time.