Integrating large language models for enhanced predictive analytics in healthcare.

Evaluating gpt-4 for zero-shot classification of bleeding and clotting events: Can large language models serve as second reviewers?

ObjectivesTo determine the extent to which current Large Language Models (LLMs) can serve as substitutes for traditional machine learning (ML) as clinical predictors using data from electronic health records (EHRs), we investigated various factors that can impact their adoption, including overall performance, calibration, fairness, and resilience to privacy protections that reduce data fidelity.Materials and MethodsWe evaluated GPT-3.5, GPT-4, and ML (as gradient-boosting trees) on clinical prediction tasks in EHR data from Vanderbilt University Medical Center and MIMIC IV. We measured predictive performance with AUROC and model calibration using Brier Score. To evaluate the impact of data privacy protections, we assessed AUROC when demographic variables are generalized. We evaluated algorithmic fairness using equalized odds and statistical parity across race, sex, and age of patients. We also considered the impact of using in-context learning by incorporating labeled examples within the prompt.ResultsTraditional ML (AUROC: 0.847, 0.894 (VUMC, MIMIC)) substantially outperformed GPT-3.5 (AUROC: 0.537, 0.517) and GPT-4 (AUROC: 0.629, 0.602) (with and without in-context learning) in predictive performance and output probability calibration (Brier Score (ML vs GPT-3.5 vs GPT-4): 0.134 versus 0.384 versus 0.251, 0.042 versus 0.06 versus 0.219). Traditional ML is more robust than GPT-3.5 and GPT-4 to generalizing demographic information to protect privacy. GPT-4 is the fairest model according to our selected metrics but at the cost of poor model performance.ConclusionThese findings suggest that LLMs are much less effective and robust than locally-trained ML for clinical prediction tasks, but they are getting better over time.

Large Language Models (LLMs) hold the potential for clinical task-shifting by processing unstructured clinical text, enabling tasks such as clinical concept extraction and medical question answering from electronic health records. If implemented reliably, such approaches may benefit over-burdened healthcare systems, particularly in resource-limited settings and for traditionally overlooked populations, provided that local fine-tuning is supported by appropriate clinical and technical expertise. However, this powerful technology remains largely understudied in real-world contexts, particularly in the Global South. This study aims to assess whether openly available LLMs can be used reliably for processing medical notes in real-world settings in South Asia. We used publicly available LLMs to parse de-identified clinical notes from a large electronic health records (EHR) database in Pakistan, containing hospital records for 8.2 million patients. ChatGPT (GPT-3.5) as a general-purpose LLM, and GatorTron (base), BioMegatron, BioBert and ClinicalBERT as medical LLMs were evaluated when applied to these data, after fine-tuning them with (a) publicly available clinical datasets namely Informatics for Integrating Biology & the Bedside (I2B2) and National NLP Clinical Challenges (N2C2) for medical concept extraction (MCE) and emrQA for medical question answering (MQA), and (b) the local Pakistani de-identified EHR dataset, which includes inpatient Discharge Summaries (DS) and Subjective, Objective, Assessment, and Plan (SOAP) notes, as detailed in this paper. MCE models were applied to these clinical notes using both 3-label and 9-label formats, while MQA models were applied to medical questions. Internal and external validation performance was measured for (a) and (b) using F1 score, precision, recall, and accuracy for MCE and BLEU and ROUGE-L, which measure lexical and sequence similarity, for MQA. When clinical LLMs were not fine-tuned on the local EHR dataset, their performance during external validation on local data was notably poorer compared to internal validation on the dataset used for fine-tuning, with reductions of at least 15% in F1 scores for MCE and 35% in ROUGE-L and BLEU scores for MQA tasks. This suggests potential bias and highlights the inability of the medical LLMs to reliably handle the data distribution of the local population without further fine-tuning and adaptation. This trend persisted across two distinct natural language processing tasks: concept extraction and question answering, spanning a spectrum of task complexities. However, fine-tuning the LLMs with local EHR data significantly improved model performance across both tasks, yielding a 7.5% to 15% increase in the F1 score for MCE and a 27% to 53% increase in ROUGE-L and BLEU scores for MQA. Notably, ChatGPT, as a general-purpose LLM, stood out as an exception, demonstrating superior performance across all measured metrics on the local dataset compared to the publicly available dataset, with improvements ranging from 3% to 17% on the local EHR dataset, even without fine-tuning on the local data. Publicly available LLMs, predominantly trained on data from high-income regions, were found to be unreliable when applied in a real-world clinical setting in Pakistan. Fine-tuning them with local EHR data and regional clinical contexts improved their reliability, demonstrating a feasible adaptation strategy that is substantially less resource-intensive than training large language models from scratch. Close collaboration between local clinical and technical experts to curate and leverage more representative, inclusive, and unbiased medical datasets, can play a crucial role in further ensuring reliability of LLMs for resource-limited, overburdened settings, to be used in ways that are safe, fair, and beneficial for all.

Large language models for electronic health records in pediatric and surgical care: A systematic review.

Electronic health records (EHRs) with clinical decision support tools are now ubiquitous in healthcare organizations. Clinical foundation models (CFMs) pretrained on large-scale, heterogeneous structured EHR data have emerged as a powerful approach to improve predictive performance and generalizability. Meanwhile, large language models (LLMs) pretrained on broad data sources are being applied to an expanding range of healthcare tasks. However, it remains unclear whether generalist LLMs can match specialized CFMs for disease risk prediction using structured clinical data. We compared CFMs (Med-BERT, CLMBR) against fine-tuned generalist LLMs (Mistral, LLaMA-2/3/3.1), a clinical LLM (Me-LLaMA), and LLM-generated embeddings paired with simple classifiers (using DeepSeek, Qwen3, and GPT-OSS) on two disease risk prediction tasks: heart failure risk among diabetic patients (DHF) and pancreatic cancer diagnosis (PaCa). Evaluations spanned multi-site EHR data, claims data, and an open-source single-institution benchmark (EHRSHOT). Performance was assessed using the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). On larger EHR and claims cohorts (>30,000 patients), fine-tuned CFMs outperformed fine-tuned LLMs by a small but statistically significant margin (<1% AUROC). The clinical LLM performed comparably to generalist LLMs despite being smaller. On the open-source PaCa cohort (3,810 patients, 199 cases), LLMs achieved slightly higher AUROCs that were not statistically significant (LLaMA-3.1-70B 86.1% vs. Med-BERT 85.3%, p=0.27), but CFMs achieved significantly higher AUPRC (Med-BERT 55.9% vs. LLaMA-3.1-70B 41.1%, p=0.001). Notably, LLM-generated trajectory embeddings paired with logistic regression or a simple MLP, without any LLM fine-tuning, achieved the best overall performance, with AUROC exceeding 90% (Qwen3) and AUPRC reaching 66% (GPT-OSS 20B). LLM-generated embeddings with lightweight classifiers outperformed both fine-tuned CFMs and fine-tuned LLMs on AUROC and AUPRC. While these results demonstrate the potential of generalist models to match or surpass specialized CFMs, their substantially greater computational cost and variable AUPRC performance in the fine-tuning setting warrant caution. We provide a reproducible evaluation framework and codebase to support continued benchmarking.

e23161 Background: Data retrieval is challenging in clinical research and traditional methods for data collection are often time-consuming and may be error-prone. Large Language Models (LLMs) have shown zero-shot capabilities in converting unstructured clinical text into structured data. These technologies could support the retrieval stage of clinical trials by leveraging the information reported in Electronic Health Records (EHRs) without relying any longer on manual curation. APOLLO 11 Consortium (NCT05550961) is a multicentric Italian trial which leverages a federated infrastructure for the analysis of advanced lung cancer patient data across Italy. Methods: We conducted a pilot study using Llama 3.1 8B on 358 Non-Small Cell Lung Cancer patients from the IRCCS Istituto Nazionale dei Tumori, leader of the APOLLO 11 Consortium. Anonymized EHRs have been analyzed within the LLM pipeline for feature extraction by Wiest et al. A combination of zero/few shot prompting techniques both in English and Italian languages was used. We selected smoking, histology, PD-L1 and staging as multiclass variables and bone/brain/liver metastases as binary variables. The ground truth collection involved a first Manual Data Entry (1-MDE) and a final full-revised MDE (2-MDE). The LLM accuracy was calculated only for the comparison LLM vs 2-MDE. In addition, we calculated the percentage of Missing Information (% MI) in 1-MDE, 2-MDE and LLM extraction. Results: Compared to 2-MDE, LLM achieved feature-specific accuracies of 0.78 for PD-L1, 0.85 for BONE METASTASIS, 0.83 for BRAIN METASTASIS, 0.89 for LIVER METASTASIS and 0.96 for TUMOUR STAGING. For smoking and staging, LLM extraction also reduced % MI relative to 1-MDE (Table 1). Only for PD-L1, we further analyzed the 12.8% of MI and found that 91.3% resulted from hallucinations (i.e., PD-L1 was misclassified as missing). Evaluations using English prompts confirmed the pipeline’s adaptability and high tasks accuracy. Conclusions: This study confirms the feasibility of LLMs for data retrieval in clinical trials demonstrating strong performance across diverse clinical features with minimal prompt optimization. LLMs could assist clinicians and data entry personnel in the 1-MDE process, streamlining initial data structuring and saving time. The 2-MDE step can remain as a quality check to address any discrepancies. Further improvements could focus on prompt optimization and integrating human feedback to reduce hallucination rates. Clinical trial information: NCT05550961 . %MI in 1-MDE, 2-MDE and LLM extraction. Accuracy refers only to LLM vs 2-MDE. Histology and metastasis sites were collected only in 2-MDE. NA = not available. Smoking PD-L1 Histology Bone Met Brain Met Liver Met T N M Stage % MI 1-MDE 6.4 8.9 NA NA NA NA 22.5 22.5 23.11 98.3 % MI 2-MDE 6.6 3 0 0 0 0 0 0 0 0 % MI LLM 2.7 12.8 10.3 0 0 0 0 0 0 6.9 % accuracy (LLM vs 2-MDE) 67 78 91 85 83 89 39 52 70 96

SummarySustained engagement in HIV care and adherence to antiretroviral therapy (ART) are essential for achieving the UNAIDS “95-95-95” targets. Despite increased ART access, disengagement from care remains a significant issue, particularly in sub-Saharan Africa. Traditional machine learning (ML) models have shown moderate success in predicting care disengagement, which would enable early intervention. We develop an enhanced large language model (LLM) fine-tuned with electronic medical records (EMRs) to predict people at risk of disengaging from HIV care in Tanzania and to provide interpretative insights into modifiable risk factors.MethodsWe developed a novel AI model by enhancing a pre-trained LLM (LLaMA 3.1, an open-source pre-trained LLM released by Meta) using routinely collected EMRs from Tanzania’s National HIV Care and Treatment Program from January 1, 2018, to June 30, 2023 (4,809,765 records for 261,192 people) to identify people at risk of disengaging from HIV care or developing adverse outcomes. Outcomes included risk of ART non-adherence, non-suppressed viral load, and loss to follow-up. Models were evaluated internally (Kagera region) and externally (Geita region), with performance compared against state-of-art ML models and zero-shot LLMs. Additionally, a team of HIV physicians in Tanzania assessed the LLM’s predictions along with LLM provided justifications for a subset of patient records to evaluate their clinical relevance and reasoning.FindingsThe enhanced LLMs consistently outperformed the supervised ML model and zero-shot LLMs across all outcomes in both internal and external validation datasets. When focusing on the 25% of PLHIV predicted as most likely to lost-to-follow-up (LTFU), the model correctly identified 78% (2,515 of 3,224) of people living with HIV (PLHIV) genuinely at risk in internal validation and 73% (7,105 of 9,733) in external validation. Attention score analysis indicated that the enhanced LLM focused on keywords such as gaps in follow-up care and ART adherence. The human expert evaluation showed alignment between clinician assessments and the LLM’s predictions in 65% of cases, with experts finding the model’s justifications reasonable and clinically relevant in 92.3% of aligned cases.InterpretationIf implemented in HIV clinics, this LLM-based AI model could help allocate resources efficiently and deliver targeted interventions, improving retention in care and advancing the UNAIDS “95-95-95” targets. By functioning like a clinician–analyzing patient summaries, predicting risks, and offering reasoning–the enhanced LLM could be integrated into clinical workflows to complement human expertise, facilitating timely interventions and informed decision-making. If implemented widely, this human-AI collaboration has the potential to improve health outcomes for people living with HIV and reduce onward transmission.FundingThe study was supported by a grant from the US National Institutes of Health (NIH): NIH NIMH 1R01MH125746.

Large language models (LLMs) have shown promising capabilities across diverse domains, yet their application to complex clinical prediction tasks remains limited. In this study, we present CARE-AD (Collaborative Analysis and Risk Evaluation for Alzheimer's Disease), a multi-agent LLM-based framework for forecasting Alzheimer's disease (AD) onset by analyzing longitudinal electronic health record (EHR) notes. CARE-AD assigns specialized LLM agents to extract signs and symptoms relevant to AD and conduct domain-specific evaluations-emulating a collaborative diagnostic process. In a retrospective evaluation, CARE-AD achieved higher accuracy (0.53 vs. 0.26-0.45) than baseline single-model approaches in predicting AD risk 10 years prior to the first recorded diagnosis code. These findings highlight the feasibility of using multi-agent LLM systems to support early risk assessment for AD and motivate further research on their integration into clinical decision support workflows.

Clinical predictive analysis is a crucial task with numerous applications and has been extensively studied using machine learning approaches. Clinical notes, a vital data source, have been employed to develop natural language processing (NLP) models for risk prediction in healthcare with robust performance. However, clinical notes vary considerably in text composition-written by diverse healthcare providers for different purposes-and the impact of these variations on NLP modeling is also underexplored. It also remains uncertain whether the recent Large Language Models (LLMs) with instruction-following capabilities can effectively handle the risk prediction task out-of-the-box, especially when using routinely collected clinical notes instead of polished text. We address these two important research questions in the context of in-hospital mortality prediction within the critical care setting. Specifically, we propose a supervised hierarchical network with note-specific modules to account for variations across different note categories, and provide a detailed comparison with strong supervised baselines and LLMs. We benchmark 34 instruction-following LLMs based on zero-shot, few-shot, and chain-of-thought prompting with diverse prompt templates. Our results demonstrate that the note-specific network delivers improved risk prediction performance compared to established supervised baselines from both measurement-based and text-based modeling. In contrast, LLMs consistently underperform on this critical task, despite their remarkable performances in other domains. This highlights important limitations and raises caution regarding the use of LLMs for risk assessment in the critical setting. Additionally, we show that the proposed model can be leveraged to select informative clinical notes to enhance the training of other models.

Assessing opioid overdose risk: a review of clinical prediction models utilizing patient-level data

Accurate interpretation of clinical test results is essential for diagnosing and managing complex immunological disorders. We explored the potential of large language models (LLMs) to automate interpretative reports for immunodeficiency and immune competence assessment via quantitative lymphocyte profiling and B-cell subset phenotyping (QLP/BSP). An LLM was fine-tuned using parameter-efficient techniques on a dataset consisting of immunophenotyping test results and corresponding interpretive reports. Model performance was compared against a retrieval-based method and expert pathologist reports. A novel automated evaluation framework assessed both the accuracy of cell population comments and the clinical relevance of generated interpretations. A custom application was built to simulate clinical workflows and measure the impact on pathologist efficiency and accuracy. Fine-tuned LLMs achieved accuracy comparable to expert pathologists in identifying and commenting on abnormal cell type counts and frequencies, in comparison the retrieval-based method exhibited substantial error rates. There was no significant difference between the rate at which abnormalities for cell subtypes were commented on between the LLMs and pathologists. More importantly, LLMs significantly reduced the time required for a single pathologist to finalize reports with a mean reduction in time of 29%. Our results suggest that LLMs hold promise for enhancing efficiency and consistency in the clinical laboratory setting. By automating aspects of interpretive reporting, LLMs can potentially reduce pathologist workload and improve the turnaround time for critical diagnostic information, while requiring expert pathologist oversight.

Leveraging heterogeneous tabular of EHRs with prompt learning for clinical prediction.

Background: Accurate identification of cancer progression events from electronic health records (EHRs) can help enable promising oncology applications such as predicting disease trajectory, assessing treatment efficacy, and generating real-world evidence. These use cases require both large-scale and high-quality data but manual abstraction of real-world progression (rwP) is time-intensive, difficult to scale, and inherently challenging given the varied and unstructured ways it can be documented across cancer types. Large language models (LLMs) offer a scalable alternative, but their accuracy relative to expert human abstractors is unclear. We evaluated the ability of LLMs to extract rwP events and dates across 7 cancer types and assessed how using LLM-extracted data impacted real-world progression-free survival (rwPFS) estimates compared to using human-abstracted data. Methods: We applied LLM-based extraction techniques to unstructured EHR text for 7 cancer types from the Flatiron Health Research Database: bladder (N=377), breast (N=1000), colorectal (N=564), hepatocellular (N=217), renal cell (N= 229), non–small cell lung (N=1000), and small cell lung (N=955). Prompt engineering strategies including zero-shot, few-shot, and chain-of-thought were tested to optimize performance. We measured agreement between the LLM and abstractor on the presence of rwP (Y/N) and first rwP date (± 30 days) in the first-line (1L) setting. To contextualize the LLM’s ability to extract rwP relative to an expert human abstractor, we evaluated the difference in F1 scores between both curation approaches using a duplicate human-abstracted reference dataset. We also compared rwPFS calculated from LLM-curated data versus human abstractor-curated data for 1000 patients in each cancer type, indexed to 1L start date. Results: Across all cancer types, agreement between the LLM and abstractor on the presence of at least 1 rwP event ranged from 86%-90% while first rwP date agreement ranged from 80%-92%. The difference in F1 score between the LLM and human abstraction was within 3-8 points across cancer types. A comparison of rwPFS between the LLM and human abstractors showed &amp;lt;1 month difference in median rwPFS and overlapping 95% confidence intervals across all cancer types. Discussion: LLMs extracted rwP with high performance, achieving F1 scores similar to expert human abstraction. Across 7 distinct cancers, agreement with human-abstracted data aligned with published inter-abstractor reliability benchmarks, and rwPFS estimates were nearly identical across curation approaches, demonstrating both the generalizability and validity of the approach. These results highlight the potential of LLMs to extract high-quality clinical endpoints at scale, helping to advance research, enhance applications such as predictive algorithms, and ultimately supporting more personalized and effective cancer care. Citation Format: Aaron B. Cohen, Konstantin Krismer, Kelly Magee, James Gippetti, Aaron Dolor, Tori Williams, Erin Fidyk, Hank Kim, Qianyu Yuan, Melissa Estevez. Using large language models for scalable extraction of real-world progression events across multiple cancer types [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr B006.

The emergence of large language models (LLMs) and the increasing prevalence of electronic health records (EHRs) present significant opportunities for advancing health care research and practice. However, research that compares and applies LLMs to extract key epilepsy-related information from unstructured medical free text is under-explored. This study fills this gap by comparing and applying different open-source LLMs and methods to extract epilepsy information from unstructured clinic letters, thereby optimizing EHRs as a resource for the benefit of epilepsy research. We also highlight some limitations of LLMs. Employing a dataset of 280 annotated clinic letters from King's College Hospital, we explored the efficacy of open-source LLMs (Llama and Mistral series) for extracting key epilepsy-related information, including epilepsy type, seizure type, current anti-seizure medications (ASMs), and associated symptoms. The study used various extraction methods, including direct extraction, summarized extraction, and contextualized extraction, complemented by role-prompting and few-shot prompting techniques. Performance was evaluated against a gold standard dataset, and was also compared to advanced fine-tuned models and human annotations. Llama 2 13b (a 13-billion-parameter LLM developed by Meta) demonstrated superior extraction capabilities across tasks by consistently outperforming other LLMs (F1 = .80 in epilepsy-type extraction, F1 = .76 in seizure-type extraction, and F1 = .90 in current ASMs extraction). Here, F1 score is a balanced metric indicating the model's accuracy in correctly identifying relevant information without excessive false positives. The study highlights the direct extraction showing consistent high performance. Comparative analysis showed that LLMs outperformed current approaches like MedCAT (Medical Concept Annotation Tool) in extracting epilepsy-related information (.2 higher in F1). The results affirm the potential of LLMs in medical information extraction relating to epilepsy, offering insights into leveraging these models for detailed and accurate data extraction from unstructured texts. The study underscores the importance of method selection in optimizing extraction performance and suggests a promising avenue for enhancing medical research and patient care through advanced natural language processing technologies.

Clinical diagnosis prediction models, when provided with a patient's medical history, aim to detect potential diseases early, facilitating timely intervention and improving prognostic outcomes. However, the inherent scarcity of patient data and large disease candidate space often pose challenges in developing satisfactory models for this intricate task. The exploration of leveraging Large Language Models (LLMs) for encapsulating clinical decision processes has been limited. We introduce MERA, a clinical diagnosis prediction model that bridges pertaining natural language knowledge with medical practice. We apply hierarchical contrastive learning on a disease candidate ranking list to alleviate the large decision space issue. With concept memorization through fine-tuning, we bridge the natural language clinical knowledge with medical codes. Experimental results on MIMIC-III and IV datasets show that MERA achieves the state-of-the-art diagnosis prediction performance and dramatically elevates the diagnosis prediction capabilities of generative LMs.