Richard Wyss

Jiageng Wu, Bowen Gu, Ren Zhou et al. · harvard, mit

Large language models (LLMs) hold great promise for medical applications and are evolving rapidly, with new models being released at an accelerated pace. However, benchmarking on large-scale real-world data such as electronic health records (EHRs) is critical, as clinical decisions are directly informed by these sources, yet current evaluations remain limited. Most existing benchmarks rely on medical exam-style questions or PubMed-derived text, failing to capture the complexity of real-world clinical data. Others focus narrowly on specific application scenarios, limiting their generalizability across broader clinical use. To address this gap, we present BRIDGE, a comprehensive multilingual benchmark comprising 87 tasks sourced from real-world clinical data sources across nine languages. It covers eight major task types spanning the entire continuum of patient care across six clinical stages and 20 representative applications, including triage and referral, consultation, information extraction, diagnosis, prognosis, and billing coding, and involves 14 clinical specialties. We systematically evaluated 95 LLMs (including DeepSeek-R1, GPT-4o, Gemini series, and Qwen3 series) under various inference strategies. Our results reveal substantial performance variation across model sizes, languages, natural language processing tasks, and clinical specialties. Notably, we demonstrate that open-source LLMs can achieve performance comparable to proprietary models, while medically fine-tuned LLMs based on older architectures often underperform versus updated general-purpose models. The BRIDGE and its corresponding leaderboard serve as a foundational resource and a unique reference for the development and evaluation of new LLMs in real-world clinical text understanding. The BRIDGE leaderboard: https://huggingface.co/spaces/YLab-Open/BRIDGE-Medical-Leaderboard

Chong Shao, Douglas Snyder, Chiran Li et al. · harvard

Identifying medication discontinuations in electronic health records (EHRs) is vital for patient safety but is often hindered by information being buried in unstructured notes. This study aims to evaluate the capabilities of advanced open-sourced and proprietary large language models (LLMs) in extracting medications and classifying their medication status from EHR notes, focusing on their scalability on medication information extraction without human annotation. We collected three EHR datasets from diverse sources to build the evaluation benchmark. We evaluated 12 advanced LLMs and explored multiple LLM prompting strategies. Performance on medication extraction, medication status classification, and their joint task (extraction then classification) was systematically compared across all experiments. We found that LLMs showed promising performance on the medication extraction and discontinuation classification from EHR notes. GPT-4o consistently achieved the highest average F1 scores in all tasks under zero-shot setting - 94.0% for medication extraction, 78.1% for discontinuation classification, and 72.7% for the joint task. Open-sourced models followed closely, Llama-3.1-70B-Instruct achieved the highest performance in medication status classification on the MIV-Med dataset (68.7%) and in the joint task on both the Re-CASI (76.2%) and MIV-Med (60.2%) datasets. Medical-specific LLMs demonstrated lower performance compared to advanced general-domain LLMs. Few-shot learning generally improved performance, while CoT reasoning showed inconsistent gains. LLMs demonstrate strong potential for medication extraction and discontinuation identification on EHR notes, with open-sourced models offering scalable alternatives to proprietary systems and few-shot can further improve LLMs' capability.

Janick Weberpals, Pamela A. Shaw, Kueiyu Joshua Lin et al.

Multiple imputation (MI) models can be improved by including auxiliary covariates (AC), but their performance in high-dimensional data is not well understood. We aimed to develop and compare high-dimensional MI (HDMI) approaches using structured and natural language processing (NLP)-derived AC in studies with partially observed confounders. We conducted a plasmode simulation study using data from opioid vs. non-steroidal anti-inflammatory drug (NSAID) initiators (X) with observed serum creatinine labs (Z2) and time-to-acute kidney injury as outcome. We simulated 100 cohorts with a null treatment effect, including X, Z2, atrial fibrillation (U), and 13 other investigator-derived confounders (Z1) in the outcome generation. We then imposed missingness (MZ2) on 50% of Z2 measurements as a function of Z2 and U and created different HDMI candidate AC using structured and NLP-derived features. We mimicked scenarios where U was unobserved by omitting it from all AC candidate sets. Using LASSO, we data-adaptively selected HDMI covariates associated with Z2 and MZ2 for MI, and with U to include in propensity score models. The treatment effect was estimated following propensity score matching in MI datasets and we benchmarked HDMI approaches against a baseline imputation and complete case analysis with Z1 only. HDMI using claims data showed the lowest bias (0.072). Combining claims and sentence embeddings led to an improvement in the efficiency displaying the lowest root-mean-squared-error (0.173) and coverage (94%). NLP-derived AC alone did not perform better than baseline MI. HDMI approaches may decrease bias in studies with partially observed confounders where missingness depends on unobserved factors.

Cheng Ju, Richard Wyss, Jessica M. Franklin et al.

Propensity score (PS) based estimators are increasingly used for causal inference in observational studies. However, model selection for PS estimation in high-dimensional data has received little attention. In these settings, PS models have traditionally been selected based on the goodness-of-fit for the treatment mechanism itself, without consideration of the causal parameter of interest. Collaborative minimum loss-based estimation (C-TMLE) is a novel methodology for causal inference that takes into account information on the causal parameter of interest when selecting a PS model. This "collaborative learning" considers variable associations with both treatment and outcome when selecting a PS model in order to minimize a bias-variance trade off in the estimated treatment effect. In this study, we introduce a novel approach for collaborative model selection when using the LASSO estimator for PS estimation in high-dimensional covariate settings. To demonstrate the importance of selecting the PS model collaboratively, we designed quasi-experiments based on a real electronic healthcare database, where only the potential outcomes were manually generated, and the treatment and baseline covariates remained unchanged. Results showed that the C-TMLE algorithm outperformed other competing estimators for both point estimation and confidence interval coverage. In addition, the PS model selected by C-TMLE could be applied to other PS-based estimators, which also resulted in substantive improvement for both point estimation and confidence interval coverage. We illustrate the discussed concepts through an empirical example comparing the effects of non-selective nonsteroidal anti-inflammatory drugs with selective COX-2 inhibitors on gastrointestinal complications in a population of Medicare beneficiaries.

Cheng Ju, Mary Combs, Samuel D Lendle et al.

The optimal learner for prediction modeling varies depending on the underlying data-generating distribution. Super Learner (SL) is a generic ensemble learning algorithm that uses cross-validation to select among a "library" of candidate prediction models. The SL is not restricted to a single prediction model, but uses the strengths of a variety of learning algorithms to adapt to different databases. While the SL has been shown to perform well in a number of settings, it has not been thoroughly evaluated in large electronic healthcare databases that are common in pharmacoepidemiology and comparative effectiveness research. In this study, we applied and evaluated the performance of the SL in its ability to predict treatment assignment using three electronic healthcare databases. We considered a library of algorithms that consisted of both nonparametric and parametric models. We also considered a novel strategy for prediction modeling that combines the SL with the high-dimensional propensity score (hdPS) variable selection algorithm. Predictive performance was assessed using three metrics: the negative log-likelihood, area under the curve (AUC), and time complexity. Results showed that the best individual algorithm, in terms of predictive performance, varied across datasets. The SL was able to adapt to the given dataset and optimize predictive performance relative to any individual learner. Combining the SL with the hdPS was the most consistent prediction method and may be promising for PS estimation and prediction modeling in electronic healthcare databases.