The paper highlights hidden data challenges in clinical machine learning, focusing on how commonly reused preprocessing pipelines for electronic health records embed assumptions about data availability, episode definitions, and missingness. Using MIMIC-IV, the authors identify violations in temporal, semantic, and population alignment—leading to temporal leakage, feature-task misalignment, and biased imputation—and show that these ambiguities can alter model signals without affecting overall predictive performance. They provide practical recommendations and open-source code for more temporally valid, clinically aligned, and reproducible preprocessing.
arXiv:2606. 17106v1 Announce Type: new Abstract: Laboratory tests in electronic health records are collected irregularly, and the absence of a test order can be as informative as the measurement itself.
arXiv:2607. 05613v1 Announce Type: new Abstract: Clinical care often relies on key laboratory indicators, yet real-world patient visits are sparse and tests are ordered irregularly, leading to pervasive missingness.
arXiv:2505. 16941v4 Announce Type: replace-cross Abstract: Foundation models (FMs) promise to address core limitations of traditional supervised machine learning: (i) reliance on large amounts of labeled data, (ii) task specificity, and (iii) poor transportability.
arXiv:2608.29582v1 Announce Type: cross Abstract: Current evaluations of large language models (LLMs) primarily focus on factual knowledge retrieval, overlooking the fundamental challenge of navigati...
arXiv:2604.04614v3 Announce Type: replace Abstract: Deep learning-based modeling of multimodal Electronic Health Records (EHRs) has become an important approach for clinical diagnosis and risk predic...
arXiv:2608. 12805v1 Announce Type: new Abstract: Access to clinical data is essential for developing reliable healthcare machine learning systems, but direct use of electronic health records is constrained by privacy regulation, institutional review, data-use agreements, and the risk of re-identification.
Fed-ReMasker is a federated learning approach that adapts the ReMasker masked autoencoder for tabular data imputation, specifically addressing feature-level missingness where entire features are absent at some centers. The method enables centers to impute unobserved features by leveraging knowledge from collaborating institutions. In benchmark tests on synthetic and real-world datasets, Fed-ReMasker achieves the lowest imputation error in the majority of scenarios and remains robust to client heterogeneity, closely matching the performance of a centralized model.
arXiv:2602. 12542v2 Announce Type: replace-cross Abstract: Deep learning models for clinical event prediction on electronic health records (EHR) often suffer performance degradation when deployed under different data distributions.
arXiv:2607. 06163v1 Announce Type: cross Abstract: Foundation Models for Electronic Health Records (FEMRs) are pretrained on large-scale structured patient data, enabling them to convert longitudinal patient trajectories into generalizable representations for diverse clinical prediction tasks.
arXiv:2609.15713v1 Announce Type: new Abstract: Recent approaches to 30-day hospital readmission prediction rely on pre-trained language models applied to discharge summaries. Although these methods...
The paper introduces Scaling Electronic Health Record Foundation Models for Population Health Management, a large‑scale model trained on billions of medical events from over 5 million patients in Taiwan and the United States. By aligning ICD codes across different health systems, the model achieves strong scaling and generalization across 11 chronic disease prediction tasks, outperforming tree‑based, general, and biomedical language models with high sensitivity at 99% specificity. It also demonstrates superior few‑shot performance on the EHRShot benchmark and shows that cross‑system alignment provides a stronger pretraining signal than single‑site duplication in data‑limited scenarios.
The paper introduces a pre‑training pipeline that creates transformer‑based imputation specialists for tabular data with specific missingness patterns. By featurizing entries, generating synthetic data with configurable missingness modules, and fitting on millions of synthetic tables, the pipeline produces pattern‑specific models that outperform dedicated methods for each missingness pattern. A default model trained only on MCAR data, TabImpute, remains robust across all tested patterns, and the authors release the pipeline, models, and a new benchmark of 42 datasets and 11 missingness patterns.