Electronic Health Records (EHRs) offer considerable potential for clinical prediction, but their complexity and heterogeneity present significant challenges for traditional machine learning methods. Recently, domain-specific EHR foundation models trained on large volumes of unlabeled EHR data have shown improved predictive accuracy and generalization. However, their development is constrained by limited access to diverse, high-quality datasets, and by inconsistencies in coding standards and clinical practices. In this study, we explore the use of general-purpose Large Language Models (LLMs) to encode EHR into high-dimensional representations for downstream clinical prediction tasks. We convert structured EHR data into markdown-formatted plain text documents by replacing medical codes with natural language descriptions. This enables the use of LLMs and their extensive semantic understanding and generalization capabilities as effective encoders of EHRs without requiring access to private medical training data. We show that LLM-based embeddings can often match or even surpass the performance of a specialized EHR foundation model, CLMBR-T-Base, across 15 diverse clinical tasks from the EHRSHOT benchmark. To demonstrate generalizability, we further evaluate the approach on the UK Biobank (UKB) cohort, a population distinct from that used to train CLMBR-T-Base. Notably, one of the tested LLM-based models achieves superior performance for disease onset, hospitalization, and mortality prediction, highlighting robustness to shifts in patient populations. Our findings suggest that repurposed general-purpose LLMs for EHR encoding provide a scalable and generalizable alternative to domain-specific models for clinical prediction.
Chest X-ray (CXR) is the most frequently ordered imaging test, supporting diverse clinical tasks from thoracic disease detection to postoperative monitoring. However, task-specific classification models are limited in scope, require costly labeled data, and lack generalizability to out-of-distribution datasets. To address these challenges, we introduce CheXFound, a self-supervised vision foundation model that learns robust CXR representations and generalizes effectively across a wide range of downstream tasks. We pretrain CheXFound on a curated CXR-1M dataset, comprising over one million unique CXRs from publicly available sources. We propose a Global and Local Representations Integration (GLoRI) module for downstream adaptations, by incorporating disease-specific local features with global image features for enhanced performance in multilabel classification. Our experimental results show that CheXFound outperforms state-of-the-art models in classifying 40 disease findings across different prevalence levels on the CXR-LT 24 dataset and exhibits superior label efficiency on downstream tasks with limited training data. Additionally, CheXFound achieved significant improvements on new tasks with out-of-distribution datasets, including opportunistic cardiovascular disease risk estimation and mortality prediction. These results highlight CheXFound's strong generalization capabilities, enabling diverse adaptations with improved label efficiency. The project source code is publicly available at https://github.com/RPIDIAL/CheXFound.
The task of predicting long-term patient outcomes using supervised machine learning is a challenging one, in part because of the high variance of each patient's trajectory, which can result in the model over-fitting to the training data. Temporal difference (TD) learning, a common reinforcement learning technique, may reduce variance by generalising learning to the pattern of state transitions rather than terminal outcomes. However, in healthcare this method requires several strong assumptions about patient states, and there appears to be limited literature evaluating the performance of TD learning against traditional supervised learning methods for long-term health outcome prediction tasks. In this study, we define a framework for applying TD learning to real-time irregularly sampled time series data using a Semi-Markov Reward Process. We evaluate the model framework in predicting intensive care mortality and show that TD learning under this framework can result in improved model robustness compared to standard supervised learning methods. and that this robustness is maintained even when validated on external datasets. This approach may offer a more reliable method when learning to predict patient outcomes using high-variance irregular time series data.
Accurately predicting blood glucose (BG) levels of ICU patients is critical, as both hypoglycemia (BG < 70 mg/dL) and hyperglycemia (BG > 180 mg/dL) are associated with increased morbidity and mortality. This study presents a proof-of-concept machine learning framework, the Multi-source Irregular Time-Series Transformer (MITST), designed to predict BG levels in ICU patients. In contrast to existing methods that rely heavily on manual feature engineering or utilize limited Electronic Health Record (EHR) data sources, MITST integrates diverse clinical data--including laboratory results, medications, and vital signs without predefined aggregation. The model leverages a hierarchical Transformer architecture, designed to capture interactions among features within individual timestamps, temporal dependencies across different timestamps, and semantic relationships across multiple data sources. Evaluated using the extensive eICU database (200,859 ICU stays across 208 hospitals), MITST achieves a statistically significant ( p < 0.001 ) average improvement of 1.7 percentage points (pp) in AUROC and 1.8 pp in AUPRC over a state-of-the-art random forest baseline. Crucially, for hypoglycemia--a rare but life-threatening condition--MITST increases sensitivity by 7.2 pp, potentially enabling hundreds of earlier interventions across ICU populations. The flexible architecture of MITST allows seamless integration of new data sources without retraining the entire model, enhancing its adaptability for clinical decision support. While this study focuses on predicting BG levels, we also demonstrate MITST's ability to generalize to a distinct clinical task (in-hospital mortality prediction), highlighting its potential for broader applicability in ICU settings. MITST thus offers a robust and extensible solution for analyzing complex, multi-source, irregular time-series data.
Accurate prediction of mortality in nasopharyngeal carcinoma (NPC), a complex malignancy particularly challenging in advanced stages, is crucial for optimizing treatment strategies and improving patient outcomes. However, this predictive process is often compromised by the high-dimensional and heterogeneous nature of NPC-related data, coupled with the pervasive issue of incomplete multi-modal data, manifesting as missing radiological images or incomplete diagnostic reports. Traditional machine learning approaches suffer significant performance degradation when faced with such incomplete data, as they fail to effectively handle the high-dimensionality and intricate correlations across modalities. Even advanced multi-modal learning techniques like Transformers struggle to maintain robust performance in the presence of missing modalities, as they lack specialized mechanisms to adaptively integrate and align the diverse data types, while also capturing nuanced patterns and contextual relationships within the complex NPC data. To address these problem, we introduce IMAN: an adaptive network for robust NPC mortality prediction with missing modalities.
Background: This study aimed to evaluate and compare the performance of classical machine learning models (CMLs) and large language models (LLMs) in predicting mortality associated with COVID-19 by utilizing a high-dimensional tabular dataset. Materials and Methods: We analyzed data from 9,134 COVID-19 patients collected across four hospitals. Seven CML models, including XGBoost and random forest (RF), were trained and evaluated. The structured data was converted into text for zero-shot classification by eight LLMs, including GPT-4 and Mistral-7b. Additionally, Mistral-7b was fine-tuned using the QLoRA approach to enhance its predictive capabilities. Results: Among the CML models, XGBoost and RF achieved the highest accuracy, with F1 scores of 0.87 for internal validation and 0.83 for external validation. In the LLM category, GPT-4 was the top performer with an F1 score of 0.43. Fine-tuning Mistral-7b significantly improved its recall from 1% to 79%, resulting in an F1 score of 0.74, which was stable during external validation. Conclusion: While LLMs show moderate performance in zero-shot classification, fine-tuning can significantly enhance their effectiveness, potentially aligning them closer to CML models. However, CMLs still outperform LLMs in high-dimensional tabular data tasks.
Biological age scores are an emerging tool to characterize aging by estimating chronological age based on physiological biomarkers. Various scores have shown associations with aging-related outcomes. This study assessed the relation between an age score based on brain MRI images (BrainAge) and an age score based on metabolomic biomarkers (MetaboAge). We trained a federated deep learning model to estimate BrainAge in three cohorts. The federated BrainAge model yielded significantly lower error for age prediction across the cohorts than locally trained models. Harmonizing the age interval between cohorts further improved BrainAge accuracy. Subsequently, we compared BrainAge with MetaboAge using federated association and survival analyses. The results showed a small association between BrainAge and MetaboAge as well as a higher predictive value for the time to mortality of both scores combined than for the individual scores. Hence, our study suggests that both aging scores capture different aspects of the aging process.
This study introduces the Federated Medical Knowledge Injection (FEDMEKI) platform, a new benchmark designed to address the unique challenges of integrating medical knowledge into foundation models under privacy constraints. By leveraging a cross-silo federated learning approach, FEDMEKI circumvents the issues associated with centralized data collection, which is often prohibited under health regulations like the Health Insurance Portability and Accountability Act (HIPAA) in the USA. The platform is meticulously designed to handle multi-site, multi-modal, and multi-task medical data, which includes 7 medical modalities, including images, signals, texts, laboratory test results, vital signs, input variables, and output variables. The curated dataset to validate FEDMEKI covers 8 medical tasks, including 6 classification tasks (lung opacity detection, COVID-19 detection, electrocardiogram (ECG) abnormal detection, mortality prediction, sepsis prediction, and enlarged cardiomediastinum detection) and 2 generation tasks (medical visual question answering (MedVQA) and ECG noise clarification). This comprehensive dataset is partitioned across several clients to facilitate the decentralized training process under 16 benchmark approaches. FEDMEKI not only preserves data privacy but also enhances the capability of medical foundation models by allowing them to learn from a broader spectrum of medical knowledge without direct data exposure, thereby setting a new benchmark in the application of foundation models within the healthcare sector.
Algorithmic bias in healthcare mirrors existing data biases. However, the factors driving unfairness are not always known. Medical devices capture significant amounts of data but are prone to errors; for instance, pulse oximeters overestimate the arterial oxygen saturation of darker-skinned individuals, leading to worse outcomes. The impact of this bias in machine learning (ML) models remains unclear. This study addresses the technical challenges of quantifying the impact of medical device bias in downstream ML. Our experiments compare a "perfect world", without pulse oximetry bias, using SaO2 (blood-gas), to the "actual world", with biased measurements, using SpO2 (pulse oximetry). Under this counterfactual design, two models are trained with identical data, features, and settings, except for the method of measuring oxygen saturation: models using SaO2 are a "control" and models using SpO2 a "treatment". The blood-gas oximetry linked dataset was a suitable test-bed, containing 163,396 nearly-simultaneous SpO2 - SaO2 paired measurements, aligned with a wide array of clinical features and outcomes. We studied three classification tasks: in-hospital mortality, respiratory SOFA score in the next 24 hours, and SOFA score increase by two points. Models using SaO2 instead of SpO2 generally showed better performance. Patients with overestimation of O2 by pulse oximetry of > 3% had significant decreases in mortality prediction recall, from 0.63 to 0.59, P < 0.001. This mirrors clinical processes where biased pulse oximetry readings provide clinicians with false reassurance of patients' oxygen levels. A similar degradation happened in ML models, with pulse oximetry biases leading to more false negatives in predicting adverse outcomes.
Integrating modern machine learning and clinical decision-making has great promise for mitigating healthcare's increasing cost and complexity. We introduce the Enhanced Transformer for Health Outcome Simulation (ETHOS), a novel application of the transformer deep-learning architecture for analyzing high-dimensional, heterogeneous, and episodic health data. ETHOS is trained using Patient Health Timelines (PHTs)-detailed, tokenized records of health events-to predict future health trajectories, leveraging a zero-shot learning approach. ETHOS represents a significant advancement in foundation model development for healthcare analytics, eliminating the need for labeled data and model fine-tuning. Its ability to simulate various treatment pathways and consider patient-specific factors positions ETHOS as a tool for care optimization and addressing biases in healthcare delivery. Future developments will expand ETHOS' capabilities to incorporate a wider range of data types and data sources. Our work demonstrates a pathway toward accelerated AI development and deployment in healthcare.
Instruction-tuned Large Language Models (LLMs) can perform a wide range of tasks given natural language instructions to do so, but they are sensitive to how such instructions are phrased. This issue is especially concerning in healthcare, as clinicians are unlikely to be experienced prompt engineers and the potential consequences of inaccurate outputs are heightened in this domain. This raises a practical question: How robust are instruction-tuned LLMs to natural variations in the instructions provided for clinical NLP tasks? We collect prompts from medical doctors across a range of tasks and quantify the sensitivity of seven LLMs -- some general, others specialized -- to natural (i.e., non-adversarial) instruction phrasings. We find that performance varies substantially across all models, and that -- perhaps surprisingly -- domain-specific models explicitly trained on clinical data are especially brittle, compared to their general domain counterparts. Further, arbitrary phrasing differences can affect fairness, e.g., valid but distinct instructions for mortality prediction yield a range both in overall performance, and in terms of differences between demographic groups.
Surgery to treat elderly hip fracture patients may cause complications that can lead to early mortality. An early warning system for complications could provoke clinicians to monitor high-risk patients more carefully and address potential complications early, or inform the patient. In this work, we develop a multimodal deep-learning model for post-operative mortality prediction using pre-operative and per-operative data from elderly hip fracture patients. Specifically, we include static patient data, hip and chest images before surgery in pre-operative data, vital signals, and medications administered during surgery in per-operative data. We extract features from image modalities using ResNet and from vital signals using LSTM. Explainable model outcomes are essential for clinical applicability, therefore we compute Shapley values to explain the predictions of our multimodal black box model. We find that i) Shapley values can be used to estimate the relative contribution of each modality both locally and globally, and ii) a modified version of the chain rule can be used to propagate Shapley values through a sequence of models supporting interpretable local explanations. Our findings imply that a multimodal combination of black box models can be explained by propagating Shapley values through the model sequence.
Machine Learning (ML) has demonstrated its great potential on medical data analysis. Large datasets collected from diverse sources and settings are essential for ML models in healthcare to achieve better accuracy and generalizability. Sharing data across different healthcare institutions is challenging because of complex and varying privacy and regulatory requirements. Hence, it is hard but crucial to allow multiple parties to collaboratively train an ML model leveraging the private datasets available at each party without the need for direct sharing of those datasets or compromising the privacy of the datasets through collaboration. In this paper, we address this challenge by proposing Decentralized, Collaborative, and Privacy-preserving ML for Multi-Hospital Data (DeCaPH). It offers the following key benefits: (1) it allows different parties to collaboratively train an ML model without transferring their private datasets; (2) it safeguards patient privacy by limiting the potential privacy leakage arising from any contents shared across the parties during the training process; and (3) it facilitates the ML model training without relying on a centralized server. We demonstrate the generalizability and power of DeCaPH on three distinct tasks using real-world distributed medical datasets: patient mortality prediction using electronic health records, cell-type classification using single-cell human genomes, and pathology identification using chest radiology images. We demonstrate that the ML models trained with DeCaPH framework have an improved utility-privacy trade-off, showing it enables the models to have good performance while preserving the privacy of the training data points. In addition, the ML models trained with DeCaPH framework in general outperform those trained solely with the private datasets from individual parties, showing that DeCaPH enhances the model generalizability.
Predicting in-hospital mortality for intensive care unit (ICU) patients is key to final clinical outcomes. AI has shown advantaged accuracy but suffers from the lack of explainability. To address this issue, this paper proposes an eXplainable Multimodal Mortality Predictor (X-MMP) approaching an efficient, explainable AI solution for predicting in-hospital mortality via multimodal ICU data. We employ multimodal learning in our framework, which can receive heterogeneous inputs from clinical data and make decisions. Furthermore, we introduce an explainable method, namely Layer-Wise Propagation to Transformer, as a proper extension of the LRP method to Transformers, producing explanations over multimodal inputs and revealing the salient features attributed to prediction. Moreover, the contribution of each modality to clinical outcomes can be visualized, assisting clinicians in understanding the reasoning behind decision-making. We construct a multimodal dataset based on MIMIC-III and MIMIC-III Waveform Database Matched Subset. Comprehensive experiments on benchmark datasets demonstrate that our proposed framework can achieve reasonable interpretation with competitive prediction accuracy. In particular, our framework can be easily transferred to other clinical tasks, which facilitates the discovery of crucial factors in healthcare research.
Survival models can help medical practitioners to evaluate the prognostic importance of clinical variables to patient outcomes such as mortality or hospital readmission and subsequently design personalized treatment regimes. Electronic Health Records (EHRs) hold the promise for large-scale survival analysis based on systematically recorded clinical features for each patient. However, existing survival models either do not scale to high dimensional and multi-modal EHR data or are difficult to interpret. In this study, we present a supervised topic model called MixEHR-SurG to simultaneously integrate heterogeneous EHR data and model survival hazard. Our contributions are three-folds: (1) integrating EHR topic inference with Cox proportional hazards likelihood; (2) integrating patient-specific topic hyperparameters using the PheCode concepts such that each topic can be identified with exactly one PheCode-associated phenotype; (3) multi-modal survival topic inference. This leads to a highly interpretable survival topic model that can infer PheCode-specific phenotype topics associated with patient mortality. We evaluated MixEHR-SurG using a simulated dataset and two real-world EHR datasets: the Quebec Congenital Heart Disease (CHD) data consisting of 8,211 subjects with 75,187 outpatient claim records of 1,767 unique ICD codes; the MIMIC-III consisting of 1,458 subjects with multi-modal EHR records. Compared to the baselines, MixEHR-SurG achieved a superior dynamic AUROC for mortality prediction, with a mean AUROC score of 0.89 in the simulation dataset and a mean AUROC of 0.645 on the CHD dataset. Qualitatively, MixEHR-SurG associates severe cardiac conditions with high mortality risk among the CHD patients after the first heart failure hospitalization and critical brain injuries with increased mortality among the MIMIC-III patients after their ICU discharge.
Within the intensive care unit (ICU), a wealth of patient data, including clinical measurements and clinical notes, is readily available. This data is a valuable resource for comprehending patient health and informing medical decisions, but it also contains many challenges in analysis. Deep learning models show promise in extracting meaningful patterns, but they require extensive labeled data, a challenge in critical care. To address this, we propose a novel approach employing self-supervised pretraining, focusing on the alignment of clinical measurements and notes. Our approach combines contrastive and masked token prediction tasks during pretraining. Semi-supervised experiments on the MIMIC-III dataset demonstrate the effectiveness of our self-supervised pretraining. In downstream tasks, including in-hospital mortality prediction and phenotyping, our pretrained model outperforms baselines in settings where only a fraction of the data is labeled, emphasizing its ability to enhance ICU data analysis. Notably, our method excels in situations where very few labels are available, as evidenced by an increase in the AUC-ROC for in-hospital mortality by 0.17 and in AUC-PR for phenotyping by 0.1 when only 1% of labels are accessible. This work advances self-supervised learning in the healthcare domain, optimizing clinical insights from abundant yet challenging ICU data.
Prediction of mortality in intensive care unit (ICU) patients typically relies on black box models (that are unacceptable for use in hospitals) or hand-tuned interpretable models (that might lead to the loss in performance). We aim to bridge the gap between these two categories by building on modern interpretable ML techniques to design interpretable mortality risk scores that are as accurate as black boxes. We developed a new algorithm, GroupFasterRisk, which has several important benefits: it uses both hard and soft direct sparsity regularization, it incorporates group sparsity to allow more cohesive models, it allows for monotonicity constraint to include domain knowledge, and it produces many equally-good models, which allows domain experts to choose among them. For evaluation, we leveraged the largest existing public ICU monitoring datasets (MIMIC III and eICU). Models produced by GroupFasterRisk outperformed OASIS and SAPS II scores and performed similarly to APACHE IV/IVa while using at most a third of the parameters. For patients with sepsis/septicemia, acute myocardial infarction, heart failure, and acute kidney failure, GroupFasterRisk models outperformed OASIS and SOFA. Finally, different mortality prediction ML approaches performed better based on variables selected by GroupFasterRisk as compared to OASIS variables. GroupFasterRisk's models performed better than risk scores currently used in hospitals, and on par with black box ML models, while being orders of magnitude sparser. Because GroupFasterRisk produces a variety of risk scores, it allows design flexibility - the key enabler of practical model creation. GroupFasterRisk is a fast, accessible, and flexible procedure that allows learning a diverse set of sparse risk scores for mortality prediction.
We present a longitudinal multi-sensor dataset collected from honey bee colonies (Apis mellifera) with rich phenotypic measurements. Data were continuously collected between May-2020 and April-2021 from 53 hives located at two apiaries in Qu\'ebec, Canada. The sensor data included audio features, temperature, and relative humidity. The phenotypic measurements contained beehive population, number of brood cells (eggs, larva and pupa), Varroa destructor infestation levels, defensive and hygienic behaviors, honey yield, and winter mortality. Our study is amongst the first to provide a wide variety of phenotypic trait measurements annotated by apicultural science experts, which facilitate a broader scope of analysis. We first summarize the data collection procedure, sensor data pre-processing steps, and data composition. We then provide an overview of the phenotypic data distribution as well as a visualization of the sensor data patterns. Lastly, we showcase several hive monitoring applications based on sensor data analysis and machine learning, such as winter mortality prediction, hive population estimation, and the presence of an active and laying queen.
Effectively leveraging multimodal data such as various images, laboratory tests and clinical information is gaining traction in a variety of AI-based medical diagnosis and prognosis tasks. Most existing multi-modal techniques only focus on enhancing their performance by leveraging the differences or shared features from various modalities and fusing feature across different modalities. These approaches are generally not optimal for clinical settings, which pose the additional challenges of limited training data, as well as being rife with redundant data or noisy modality channels, leading to subpar performance. To address this gap, we study the robustness of existing methods to data redundancy and noise and propose a generalized dynamic multimodal information bottleneck framework for attaining a robust fused feature representation. Specifically, our information bottleneck module serves to filter out the task-irrelevant information and noises in the fused feature, and we further introduce a sufficiency loss to prevent dropping of task-relevant information, thus explicitly preserving the sufficiency of prediction information in the distilled feature. We validate our model on an in-house and a public COVID19 dataset for mortality prediction as well as two public biomedical datasets for diagnostic tasks. Extensive experiments show that our method surpasses the state-of-the-art and is significantly more robust, being the only method to remain performance when large-scale noisy channels exist. Our code is publicly available at https://github.com/ayanglab/DMIB.
Machine learning (ML) has recently shown promising results in medical predictions using electronic health records (EHRs). However, since ML models typically have a limited capability in terms of input sizes, selecting specific medical events from EHRs for use as input is necessary. This selection process, often relying on expert opinion, can cause bottlenecks in development. We propose Retrieval-Enhanced Medical prediction model (REMed) to address such challenges. REMed can essentially evaluate unlimited medical events, select the relevant ones, and make predictions. This allows for an unrestricted input size, eliminating the need for manual event selection. We verified these properties through experiments involving 27 clinical prediction tasks across four independent cohorts, where REMed outperformed the baselines. Notably, we found that the preferences of REMed align closely with those of medical experts. We expect our approach to significantly expedite the development of EHR prediction models by minimizing clinicians' need for manual involvement.