Collin M. Stultz

Hyewon Jeong, Collin M. Stultz, Marzyeh Ghassemi

Heart failure is a debilitating condition that affects millions of people worldwide and has a significant impact on their quality of life and mortality rates. An objective assessment of cardiac pressures remains an important method for the diagnosis and treatment prognostication for patients with heart failure. Although cardiac catheterization is the gold standard for estimating central hemodynamic pressures, it is an invasive procedure that carries inherent risks, making it a potentially dangerous procedure for some patients. Approaches that leverage non-invasive signals - such as electrocardiogram (ECG) - have the promise to make the routine estimation of cardiac pressures feasible in both inpatient and outpatient settings. Prior models trained to estimate intracardiac pressures (e.g., mean pulmonary capillary wedge pressure (mPCWP)) in a supervised fashion have shown good discriminatory ability but have been limited to the labeled dataset from the heart failure cohort. To address this issue and build a robust representation, we apply deep metric learning (DML) and propose a novel self-supervised DML with distance-based mining that improves the performance of a model with limited labels. We use a dataset that contains over 5.4 million ECGs without concomitant central pressure labels to pre-train a self-supervised DML model which showed improved classification of elevated mPCWP compared to self-supervised contrastive baselines. Additionally, the supervised DML model that uses ECGs with access to 8,172 mPCWP labels demonstrated significantly better performance on the mPCWP regression task compared to the supervised baseline. Moreover, our data suggest that DML yields models that are performant across patient subgroups, even when some patient subgroups are under-represented in the dataset. Our code is available at https://github.com/mandiehyewon/ssldml

Aniruddh Raghu, Divya Shanmugam, Eugene Pomerantsev et al.

Neural network models have demonstrated impressive performance in predicting pathologies and outcomes from the 12-lead electrocardiogram (ECG). However, these models often need to be trained with large, labelled datasets, which are not available for many predictive tasks of interest. In this work, we perform an empirical study examining whether training time data augmentation methods can be used to improve performance on such data-scarce ECG prediction problems. We investigate how data augmentation strategies impact model performance when detecting cardiac abnormalities from the ECG. Motivated by our finding that the effectiveness of existing augmentation strategies is highly task-dependent, we introduce a new method, TaskAug, which defines a flexible augmentation policy that is optimized on a per-task basis. We outline an efficient learning algorithm to do so that leverages recent work in nested optimization and implicit differentiation. In experiments, considering three datasets and eight predictive tasks, we find that TaskAug is competitive with or improves on prior work, and the learned policies shed light on what transformations are most effective for different tasks. We distill key insights from our experimental evaluation, generating a set of best practices for applying data augmentation to ECG prediction problems.

Aniruddh Raghu, Payal Chandak, Ridwan Alam et al.

Self-supervised learning (SSL) for clinical time series data has received significant attention in recent literature, since these data are highly rich and provide important information about a patient's physiological state. However, most existing SSL methods for clinical time series are limited in that they are designed for unimodal time series, such as a sequence of structured features (e.g., lab values and vitals signs) or an individual high-dimensional physiological signal (e.g., an electrocardiogram). These existing methods cannot be readily extended to model time series that exhibit multimodality, with structured features and high-dimensional data being recorded at each timestep in the sequence. In this work, we address this gap and propose a new SSL method -- Sequential Multi-Dimensional SSL -- where a SSL loss is applied both at the level of the entire sequence and at the level of the individual high-dimensional data points in the sequence in order to better capture information at both scales. Our strategy is agnostic to the specific form of loss function used at each level -- it can be contrastive, as in SimCLR, or non-contrastive, as in VICReg. We evaluate our method on two real-world clinical datasets, where the time series contains sequences of (1) high-frequency electrocardiograms and (2) structured data from lab values and vitals signs. Our experimental results indicate that pre-training with our method and then fine-tuning on downstream tasks improves performance over baselines on both datasets, and in several settings, can lead to improvements across different self-supervised loss functions.

Ridwan Alam, Collin Stultz

The diagnosis, prognosis, and treatment of a number of cardiovascular disorders rely on ECG interval measurements, including the PR, QRS, and QT intervals. These quantities are measured from the 12-lead ECG, either manually or using automated algorithms, which are readily available in clinical settings. A number of wearable devices, however, can acquire the lead-I ECG in an outpatient setting, thereby raising the potential for out-of-hospital monitoring for disorders that involve clinically significant changes in ECG intervals. In this work, we therefore developed a series of deep learning models for estimating the PR, QRS, and QT intervals using lead-I ECG. From a corpus of 4.2 million ECGs from patients at the Massachusetts General Hospital, we train and validate each of the models. At internal holdout validation, we achieve mean absolute errors (MAE) of 6.3 ms for QRS durations and 11.9 ms for QT intervals, and an MAE of 9.2 ms for estimating PR intervals. Moreover, as a well-defined P-wave does not always exist in ECG tracings - for example, when there is atrial fibrillation - we trained a model that can identify when there is a P-wave, and consequently, a measurable PR interval. We validate our models on three large external healthcare datasets without any finetuning or retraining - 3.2 million ECG from the Brigham and Womens Hospital, 668 thousand from MIMIC-IV, and 20 thousand from PTB-XL - and achieve similar performance. Also, our models significantly outperform two publicly available baseline algorithms. This work demonstrates that ECG intervals can be tracked from only lead-I ECG using deep learning, and highlights the potential for out-of-hospital applications.

Nathaniel Diamant, Erik Reinertsen, Steven Song et al.

Supervised machine learning applications in health care are often limited due to a scarcity of labeled training data. To mitigate this effect of small sample size, we introduce a pre-training approach, Patient Contrastive Learning of Representations (PCLR), which creates latent representations of ECGs from a large number of unlabeled examples. The resulting representations are expressive, performant, and practical across a wide spectrum of clinical tasks. We develop PCLR using a large health care system with over 3.2 million 12-lead ECGs, and demonstrate substantial improvements across multiple new tasks when there are fewer than 5,000 labels. We release our model to extract ECG representations at https://github.com/broadinstitute/ml4h/tree/master/model_zoo/PCLR.

Hyewon Jeong, Nassim Oufattole, Matthew Mcdermott et al.

In clinical practice, one often needs to identify whether a patient is at high risk of adverse outcomes after some key medical event. For example, quantifying the risk of adverse outcomes after an acute cardiovascular event helps healthcare providers identify those patients at the highest risk of poor outcomes; i.e., patients who benefit from invasive therapies that can lower their risk. Assessing the risk of adverse outcomes, however, is challenging due to the complexity, variability, and heterogeneity of longitudinal medical data, especially for individuals suffering from chronic diseases like heart failure. In this paper, we introduce Event-Based Contrastive Learning (EBCL) - a method for learning embeddings of heterogeneous patient data that preserves temporal information before and after key index events. We demonstrate that EBCL can be used to construct models that yield improved performance on important downstream tasks relative to other pretraining methods. We develop and test the method using a cohort of heart failure patients obtained from a large hospital network and the publicly available MIMIC-IV dataset consisting of patients in an intensive care unit at a large tertiary care center. On both cohorts, EBCL pretraining yields models that are performant with respect to a number of downstream tasks, including mortality, hospital readmission, and length of stay. In addition, unsupervised EBCL embeddings effectively cluster heart failure patients into subgroups with distinct outcomes, thereby providing information that helps identify new heart failure phenotypes. The contrastive framework around the index event can be adapted to a wide array of time-series datasets and provides information that can be used to guide personalized care.

Ridwan Alam, Aaron Aguirre, Collin Stultz

For a number of antiarrhythmics, drug loading requires a 3 day hospitalization with monitoring for QT prolongation. Automated QT monitoring with wearable ECG monitors would facilitate out-of-hospital care. We develop a deep learning model that infers QT intervals from ECG lead-I - the lead most often acquired from ambulatory ECG monitors - and to use this model to detect clinically meaningful QT-prolongation episodes during Dofetilide drug loading. Using 4.22 million 12-lead ECG recordings from 903.6 thousand patients at the Massachusetts General Hospital, we develop a deep learning model, QTNet, that infers QT intervals from lead-I. Over 3 million ECGs from 653 thousand patients are used to train the model and an internal-test set containing 633 thousand ECGs from 135 thousand patients was used for testing. QTNet is further evaluated on an external-validation set containing 3.1 million ECGs from 667 thousand patients at another institution. QTNet was used to detect Dofetilide-induced QT prolongation in a publicly available database (ECGRDVQ-dataset) containing ECGs from subjects enrolled in a clinical trial evaluating the effects of antiarrhythmic drugs. QTNet achieves mean absolute errors of 12.63ms (internal-test) and 12.30ms (external-validation) for estimating absolute QT intervals. The associated Pearson correlation coefficients are 0.91 (internal-test) and 0.92 (external-validation). For the ECGRDVQ-dataset, QTNet detects Dofetilide-induced QTc prolongation with 87% sensitivity and 77% specificity. The negative predictive value of the model is greater than 95% when the pre-test probability of drug-induced QTc prolongation is below 25%. Drug-induced QT prolongation risk can be tracked from ECG lead-I using deep learning.

Nassim Oufattole, Teya Bergamaschi, Aleksia Kolo et al. · mit

Effective, reliable, and scalable development of machine learning (ML) solutions for structured electronic health record (EHR) data requires the ability to reliably generate high-quality baseline models for diverse supervised learning tasks in an efficient and performant manner. Historically, producing such baseline models has been a largely manual effort--individual researchers would need to decide on the particular featurization and tabularization processes to apply to their individual raw, longitudinal data; and then train a supervised model over those data to produce a baseline result to compare novel methods against, all for just one task and one dataset. In this work, powered by complementary advances in core data standardization through the MEDS framework, we dramatically simplify and accelerate this process of tabularizing irregularly sampled time-series data, providing researchers the ability to automatically and scalably featurize and tabularize their longitudinal EHR data across tens of thousands of individual features, hundreds of millions of clinical events, and diverse windowing horizons and aggregation strategies, all before ultimately leveraging these tabular data to automatically produce high-caliber XGBoost baselines in a highly computationally efficient manner. This system scales to dramatically larger datasets than tabularization tools currently available to the community and enables researchers with any MEDS format dataset to immediately begin producing reliable and performant baseline prediction results on various tasks, with minimal human effort required. This system will greatly enhance the reliability, reproducibility, and ease of development of powerful ML solutions for health problems across diverse datasets and clinical settings.

Zachary Berger, Daniel Prakah-Asante, John Guttag et al.

This position paper argues that current benchmarking practice in 12-lead ECG representation learning must be fixed to ensure progress is reliable and aligned with clinically meaningful objectives. The field has largely converged on three public multi-label benchmarks (PTB-XL, CPSC2018, CSN) dominated by arrhythmia and waveform-morphology labels, even though the ECG is known to encode substantially broader clinical information. We argue that downstream evaluation should expand to include an assessment of structural heart disease and patient-level forecasting, in addition to other evolving ECG-related endpoints, as relevant clinical targets. Next, we outline evaluation best practices for multi-label, imbalanced settings, and show that when they are applied, the literature's current conclusion about which representations perform best is altered. Furthermore, we demonstrate the surprising result that a randomly initialized encoder with linear evaluation matches state-of-the-art pre-training on many tasks. This motivates the use of a random encoder as a reasonable baseline model. We substantiate our observations with an empirical evaluation of three representative ECG pre-training approaches across six evaluation settings: the three standard benchmarks, a structural disease dataset, hemodynamic inference, and patient forecasting.

Aniruddh Raghu, John Guttag, Katherine Young et al.

The impact of machine learning models on healthcare will depend on the degree of trust that healthcare professionals place in the predictions made by these models. In this paper, we present a method to provide people with clinical expertise with domain-relevant evidence about why a prediction should be trusted. We first design a probabilistic model that relates meaningful latent concepts to prediction targets and observed data. Inference of latent variables in this model corresponds to both making a prediction and providing supporting evidence for that prediction. We present a two-step process to efficiently approximate inference: (i) estimating model parameters using variational learning, and (ii) approximating maximum a posteriori estimation of latent variables in the model using a neural network, trained with an objective derived from the probabilistic model. We demonstrate the method on the task of predicting mortality risk for patients with cardiovascular disease. Specifically, using electrocardiogram and tabular data as input, we show that our approach provides appropriate domain-relevant supporting evidence for accurate predictions.

Yun Liu, Kun-Ta Chuang, Fu-Wen Liang et al.

In many domains such as medicine, training data is in short supply. In such cases, external knowledge is often helpful in building predictive models. We propose a novel method to incorporate publicly available domain expertise to build accurate models. Specifically, we use word2vec models trained on a domain-specific corpus to estimate the relevance of each feature's text description to the prediction problem. We use these relevance estimates to rescale the features, causing more important features to experience weaker regularization. We apply our method to predict the onset of five chronic diseases in the next five years in two genders and two age groups. Our rescaling approach improves the accuracy of the model, particularly when there are few positive examples. Furthermore, our method selects 60% fewer features, easing interpretation by physicians. Our method is applicable to other domains where feature and outcome descriptions are available.