BioData Mining

Very preterm infants face a considerably higher risk of neurodevelopmental delays, making early diagnosis and timely intervention crucial for improving long-term outcomes. In this study, we utilized large language models (LLMs) to predict mental and psychomotor delays at 25 months using maternal and perinatal records combined with longitudinal features up to 22 months of age. The LLMs were employed to generate natural language descriptions for each infant, which were then used as input for a language model-based classifier to perform predictions. Our model achieved a 4.2% increase in AUCROC in mental delay prediction and 3.2% increase in psychomotor delay prediction 3 months before the 25-month assessment, compared to a random forest-based model for numerical tabular data only. These findings highlight the potential of LLMs as powerful tools for assessing the risk of neurodevelopmental delays in preterm infants.

Genome-Wide Association Studies (GWAS) have successfully identified numerous loci associated with complex traits and diseases, yet pinpointing causal genes remains a significant challenge. The reliance on simple proximity-based heuristics is often insufficient due to linkage disequilibrium, gene interactions, and regulatory effects. Recent advancements in Large Language Models (LLMs) have demonstrated potential in automating causal gene identification, but their effectiveness remains limited by knowledge representation and retrieval mechanisms. This study builds on previous research by evaluating LLMs for causal gene identification, with a focus on enhancing performance through Retrieval-Augmented Generation (RAG) and the incorporation of genomic distance information. We replicate prior results using smaller model Qwen2.5--assessing their predictive accuracy using a benchmark dataset from Open Targets. We improved the preformences when integrating RAG-based literature retrieval (F1 = 0.795) and gene distance information (F1 = 0.806). However, the combined approach yielded diminishing returns, suggesting interactions between these enhancements. Error analysis revealed that genomic distance features improved predictions by reinforcing established heuristics, while RAG enhanced domain knowledge but occasionally led to semantic biases. These findings highlight the potential of hybrid approaches in leveraging both structured genomic features and unstructured textual data.

BackgroundDrug-drug interactions (DDIs) are a significant source of morbidity and adverse drug events (ADEs), particularly in situations of polypharmacy and complex medication regimens. While rules-based software integrated in electronic health records (EHRs) has demonstrated proficiency in identifying DDIs present in medication regimens, large language model (LLM) based identification requires thorough benchmarking and performance evaluation using high-quality datasets for safe use. The purpose of this study was to develop a series of performance benchmarking experiments specifically for LLM performance in identification and management of DDIs using a specifically curated clinician-annotated dataset of clinically-relevant DDIs. MethodsWe evaluated three LLMs (GPT-4o-mini, MedGemma-27B, LLaMA3-70B) using a clinician-annotated benchmark dataset of 750 DDI scenarios spanning three levels of diagnostic complexity. Tasks were aligned with flexible judgment formats: (1) a pointwise two-drug classification task, (2) a pairwise three-drug discrimination task, and (3) a listwise 4-6 drug selection task. Standardized zero-shot prompting with task-specific instructions was applied for all models. Performance was assessed using precision, recall, F1 score, and accuracy. Reliability was quantified using self-consistency across repeated runs and confidence-aligned metrics to capture stability in model reasoning. ResultsAcross the three experiments, model performance varied by task structure and interaction severity. LLaMA3-70B demonstrated the highest recall and F1 score in the pointwise task, whereas GPT-4o-mini achieved superior accuracy and consistency in the pairwise and listwise tasks. MedGemma-27B showed competitive performance in identifying Category D interactions. Self-consistency decreased as task complexity increased, highlighting reduced stability in multi-drug reasoning. No model exhibited uniformly high reliability across all judgment formats. ConclusionsCurrent LLMs show promising but uneven capabilities in identifying DDIs across clinically relevant task structures. Performance degrades as the reasoning space expands, and stability across repeated queries remains limited. These findings emphasize the need for multi-format evaluation frameworks and reliability-aware assessment when considering LLMs for medication-safety applications.

The existence of rare, genetically distinct cells can occur in various samples such as transplant patients, naturally occurring microchimerism between maternal and fetal tissues, and cancer samples with sufficient mutational burden. Computational methods for detecting these foreign cells are vital to studying these biological conditions. An application that is of particular interest is that of leukemia patients post hematopoietic cell transplant (HCT). In many leukemias, a primary therapy is HCT, after which, the primary genotype of the bone marrow and blood cells should be of donor origin. If cells exist that are of the patients genotype and the cell type lineage of the particular leukemia, this is known as measurable residual disease (MRD). If the MRD is high enough, this may represent a relapse of the patients leukemia. Furthermore, accurately estimating the MRD is important for driving clinical decision making for these patients. Here we present Cellector, a computational method for identifying rare foreign genotype cells in single cell RNAseq (scRNAseq) datasets. We show cellector accurately detects microchimeric cells down to an exceedingly low percentage of these cells present (0.05% or lower).

Background Large Language Models (LLMs) are increasingly explored for pharmacovigilance tasks, including information extraction, case documentation, and single-case causality assessment. However, their ability to support causality assessment at the case series level -- a complex, time-intensive task requiring clinical reasoning across multiple reports -- remains unexplored. Objective To investigate how a large-scale general-purpose LLM can support pharmacovigilance professionals in assessing causality in a case series, and to explore how prompt design influences the quality of the model's reasoning. Methods GPT-4o was used to assess causality for five drug - adverse event combinations, using an adaptation of the Bradford Hill viewpoints for case series assessment. The combinations represented varying drugs and vaccines, adverse events, and case series sizes (5-402 reports). One combination served as a negative control. Structured prompts were iteratively developed and refined using one combination, then applied to all combinations. LLM-generated assessments for each viewpoint were qualitatively evaluated by human annotators for accuracy (precision), and the LLM's coverage of key aspects from the original signal text was assessed for one combination (recall). Results Across all five combinations, annotators agreed with 79-92% of the LLM's output sentences. Full disagreement was consistently low (3-7%), with errors typically involving misinterpretation of complex report details rather than outright fabrication. Prompt design substantially influenced output quality; providing Bradford Hill viewpoint descriptions, including case series data, and adding explicit anti-hallucination instructions improved specificity and grounding. For the recall assessment, 15 of 23 key segments from the original signal text were reflected in the LLM output. The overall summary assessments demonstrated balanced reasoning, correctly distinguishing between positive safety signals and the negative control, and provided a coherent synthesis suitable as a starting point for human assessors. Conclusions LLMs have the potential to generate contextually nuanced and largely accurate preliminary causality assessments of case series aligned with the Bradford Hill viewpoints, with a low but non-zero hallucination rate. These findings support LLMs as a tool to augment, not replace, expert judgment in signal assessment. Future work should address larger and more diverse signal sets, improved evaluation frameworks for generative output, and the integration of pre-computed summary statistics to reduce errors.

BackgroundAcute myeloid leukemia (AML) remains the most lethal acute leukemia in adults, with 5-year overall survival below 32% despite recent advances including venetoclax-, FLT3-, IDH1/2-, and Menin-targeted therapies. Clinical outcomes remain highly heterogeneous across patients, highlighting the need for robust molecular biomarkers capable of improving prognostic precision. MicroRNAs (miRNAs) are critical regulators of hematopoietic differentiation, apoptosis, and therapeutic resistance and are differentially expressed across AML subtypes. However, their clinical translation has been limited by high dimensionality, feature redundancy, and relatively small cohort sizes. MethodsWe developed and evaluated the AML Survival Estimator (AMLS), an inheritable bi-objective combinatorial genetic algorithm integrated with support vector regression (SVR), using TCGA-LAML miRNA expression profiles (n = 156). AMLS was benchmarked against ten widely used machine-learning approaches, including penalized regression, tree-based ensembles, support-vector regression, k-nearest neighbors, and multilayer perceptron models. Performance was assessed using stratified cross-validation with Pearson correlation (R), Harrells concordance index (C-index), and mean absolute error (MAE). Functional characterization of the derived miRNA signature was performed through consensus target integration followed by pathway enrichment, gene ontology analysis, network reconstruction, and Kaplan-Meier risk stratification. ResultsAMLS achieved superior prognostic performance with pooled out-of-fold metrics of Pearson R = 0.86, C-index = 0.788, and MAE = 7.49 months, substantially outperforming all comparator models. Restricting analyses to the AMLS-derived 28-miRNA signature improved all baseline learners by approximately 2-4-fold, with the multilayer perceptron achieving R = 0.674; however, none matched the native AMLS framework, indicating that the evolutionary optimization strategy contributes predictive information beyond feature selection alone. The prognostic signature included biologically established AML-associated miRNAs, including hsa-miR-191, hsa-miR-29c, hsa-miR-125b, hsa-miR-148a, hsa-miR-15b, hsa-miR-10b, and hsa-miR-30c, linked to DNA methylation, apoptosis, cell-cycle regulation, and oncogenic Wnt/MAPK signaling pathways. Functional analyses demonstrated significant enrichment of canonical AML-associated pathways, including p53, PI3K-AKT, TGF-{beta}, JAK-STAT, FoxO, and hematopoietic lineage signaling. ConclusionsOur findings demonstrate that evolutionary learning integrated with SVR can recover a compact and biologically interpretable miRNA prognostic signature that substantially outperforms conventional machine-learning approaches for AML survival prediction. The identified miRNA network converged on key leukemogenic pathways involved in apoptosis, cell-cycle regulation, and oncogenic signaling, supporting both the biological relevance and prognostic utility of the framework. Given the minimally invasive and quantitatively scalable nature of miRNA profiling, this approach may provide a practical molecular adjunct for improving prognostic assessment and precision medicine strategies in AML. Abstract FigureSchematic overview of the AMLS framework. Left: acute myeloid leukemia, a clonal hematological malignancy with persistent prognostic heterogeneity. Middle: AMLS couples an evolutionary learning-based feature selection algorithms to support vector regression for miRNA-based survival modeling. Right: AMLS recovers a 28-miRNA prognostic signature that predicts overall survival with Pearson R = 0.86 and MAE = 7.5 months. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=86 SRC="FIGDIR/small/727196v1_ufig1.gif" ALT="Figure 1"> View larger version (20K): org.highwire.dtl.DTLVardef@11ead1org.highwire.dtl.DTLVardef@4f5c19org.highwire.dtl.DTLVardef@277de1org.highwire.dtl.DTLVardef@b95c9a_HPS_FORMAT_FIGEXP M_FIG C_FIG

Decision Curve Analysis (DCA) has emerged as a valuable method for evaluating predictive models, yet its application in research--particularly pediatrics--remains limited. This article serves as a didactic tutorial, providing an applied introduction to DCA and clarifying key concepts that often generate confusion. The advantages of DCA over traditional metrics such as the area under the ROC curve (AUC) are highlighted through a simulated cohort of pediatric patients with suspected appendicitis. Three predictors were compared: a composite clinical score (Pediatric Appendicitis Score, PAS), leukocyte count, and serum sodium. While both PAS and leukocyte count achieved acceptable AUCs, the decision curves revealed substantially different net benefit profiles, demonstrating that a higher AUC does not necessarily translate into superior clinical utility. In contrast, serum sodium, with poor discrimination, consistently failed to provide meaningful benefit across thresholds. Common methodological pitfalls--including overfitting, calibration, and the limitations of dichotomous predictors--are also discussed. By contrasting a strong predictor, a moderate predictor, and a weak biomarker, this tutorial underscores the unique contribution of DCA as a practical framework for moving beyond statistical performance toward clinically meaningful decision support.

Manual extraction of high-fidelity gene-disease-phenotype information from human genetics literature is a labor-intensive task that requires trained human genetics researchers to read through many primary research papers. This presents a major challenge for maintaining up-to-date human disease genetic databases. Recent exploration into large language models (LLMs) opens new directions in automating this manual process. However, most approaches depend on pre-training, finetuning, or specialized generative artificial intelligence (GenAI) tools, but there is a lack of empirical evidence to show whether commercially-available LLMs can be directly used to reliably extract gene-disease-phenotype for human genetic diseases. Herein, we perform a benchmark of the use of three zero-shot prompted LLMs, namely GPT-4, DeepSeek and Claude, without task-specific fine-tuning, to extract human genetic information directly from full text of scientific papers. Using known congenital heart diseases (CHD) genes found in the open access CHDgene database (https://chdgene.victorchang.edu.au/) as the benchmark data set, GPT-4o achieved overall 88.8% extraction accuracy across 23 gene entries containing over 57 references, with 100% accuracy in gene name, 78.3% and 76.7% in disease and phenotype fields respectively. This work introduces a lightweight, easy-to-deploy, and yet robust LLM-based agent named GeneAgent, analyze sources of disagreement, and highlight the feasibility of integrating powerful LLM into genetic evidence synthesis workflows. Highlight- First systematic benchmark of LLMs for extracting human gene-disease-phenotype relationships from full-text biomedical articles - GeneAgent: a lightweight, highly accurate prompt-only LLM agent - New domain task-specific evaluation framework

The practices of identifying biomarkers and developing prognostic models using genomic data has become increasingly prevalent. Such data often features characteristics that make these practices difficult, namely high dimensionality, correlations between predictors, and sparsity. Many modern methods have been developed to address these problematic characteristics while performing feature selection and prognostic modeling, but a large-scale comparison of their performances in these tasks on diverse right-censored time to event data (aka survival time data) is much needed. We have compiled many existing methods, including some machine learning methods, several which have performed well in previous benchmarks, primarily for comparison in regards to variable selection capability, and secondarily for survival time prediction on many synthetic datasets with varying levels of sparsity, correlation between predictors, and signal strength of informative predictors. For illustration, we have also performed multiple analyses on a publicly available and widely used cancer cohort from The Cancer Genome Atlas using these methods. We evaluated the methods through extensive simulation studies in terms of the false discovery rate, F1-score, concordance index, Brier score, root mean square error, and computation time. Of the methods compared, CoxBoost and the Adaptive LASSO performed well in all metrics, and the LASSO and elastic net excelled when evaluating concordance index and F1-score. The Benjamini-Hoschberg and q-value procedures showed volatile performances in controlling the false discovery rate. Some methods performances were greatly affected by differences in the data characteristics. With our extensive numerical study, we have identified the best performing methods for a plethora of data characteristics using informative metrics. This will help cancer researchers in choosing the best approach for their needs when working with genomic data.

Background: Veterans face an elevated risk of suicide compared to the general population, motivating national efforts to develop predictive models that can guide proactive care. Current models used by the U.S. Department of Veterans Affairs (VA) rely primarily on structured electronic health record (EHR) data, though clinical notes contain rich contextual information that can be quantified using natural language processing (NLP) to derive psychosocial variables that may improve risk detection. Machine learning methods, particularly classification and regression trees (CART), can also uncover interactions between clinical and psychosocial variables, enabling identification of patient characteristics that modify suicide risk factors. However, integrating structured and unstructured data presents challenges because NLP features often greatly outnumber traditional clinical variables, potentially biasing interaction discovery. In prior work, we addressed this imbalance by introducing a weighted CART framework that balances structured variables with NLP-derived psychosocial features from semantic lexicons (SEANCE). While effective, semantic approaches summarize language into predefined constructs and may overlook important lexical variation present in clinical narratives. Methods: In this study, we extend that framework by replacing semantic features with a high-dimensional bag-of-words (BoW) representation of clinical notes and by evaluating models across cohorts defined by structured suicide risk stratification (low, medium, high) and varying temporal lookback windows. Using a cohort of 27,241 veterans, we analyzed clinical documentation collected up to 30, 90, or 270 days prior to death (or a matched index date for controls), enabling temporally flexible risk modeling. XGBoost models were trained to balance structured and unstructured features and identify cross-modal interactions between textual and clinical variables. Results: When incorporated into generalized linear models, these interactions improved predictive performance, particularly among low- and medium-risk patients, and substantially reduced the performance gap between interpretable and more complex models. Notably, the BoW representation outperformed our prior semantic index-based approach. Discussion and Conclusions: Together, these findings demonstrate the utility of interpretable NLP methods for uncovering clinically meaningful interactions between psychosocial and demographic factors in suicide risk and establish a strong benchmark for future deep learning approaches aimed at capturing richer contextual and temporal information from clinical narratives.

Neonates requiring intensive care are at increased risk for long-term neuropsychiatric disorders. However, clinical adoption of risk prediction models remains limited when their performance lacks adequate interpretability for informed clinical decision-making. Here, we investigated whether longitudinal neonatal electronic health record (EHR) data from the first 90 days of life can support clinically meaningful interpretation of long-term risk signals for major neuropsychiatric diagnoses by age seven. In a retrospective register-based cohort of 17,655 at-risk children from an academic medical center, of whom 8.0\% (1,420) received a major neuropsychiatric diagnosis during follow-up, we applied a time-aware transformer model (Self-supervised Transformer for Time-Series; STraTS) and thoroughly evaluated its predictions using three complementary interpretability approaches: perturbation-based variable importance, value-dependent effect analysis, and leave-one-out (LOO) feature attribution. STraTS achieved the highest area under the precision--recall curve (AUPRC 0.171 {+/-} 0.022), compared with Random Forest (0.166 {+/-} 0.008), logistic regression (0.151 {+/-} 0.007), and XGBoost (0.128 {+/-} 0.010). Across interpretability methods, five predictors were consistently identified: birth weight, gender, Apgar score at 1 minute, umbilical serum thyroid stimulating hormone (uS-TSH), and treatment time in hospital. Indicators of early clinical severity, including chromosomal abnormalities and neonatal cerebral-status disturbances, showed the largest risk-increasing effects. Furthermore, the model's learned vector representations of subject-specific EHR sequences formed clinically coherent latent embeddings that reflect population heterogeneity along established perinatal risk dimensions. These findings demonstrate that combining multiple complementary interpretability methods yields stable, clinically plausible risk signals while revealing limitations that would remain undetected by any single approach, highlighting the importance of careful interpretability analysis of deep learning-based risk predictions.

Adverse event report (AER) data are a key source of signal for post marketing drug surveillance. The standard methodology to analyze AER data applies disproportionality metrics, which estimate the strength of drug/side-effect associations from discrete counts of their occurrence at report level. However, in other domains, improvements in predictive modeling accuracy have been obtained through representation learning, where discrete features are replaced by distributed representations learned from unlabeled data. This paper describes aer2vec, a novel representational approach for AER data in which concept embeddings emerge from neural networks trained to predict drug/side-effect co-occurrence. Trained models are evaluated for their utility in identifying drug/side-effect relationships, with improvements over disproportionality metrics in most cases. In addition, we evaluate the utility of an otherwise-untapped resource in the Food and Drug Administration (FDA) AER system - reporter designations of suspected causality - and find that incorporating this information enhances performance of all models evaluated.

AO_SCPLOWBSTRACTC_SCPLOWO_ST_ABSBackgroundC_ST_ABSLarge Language Models (LLMs) have demonstrated significant potential in clinical medicine, but a persistent performance gap exists in the pediatric domain due to its unique complexities. This study provides the first comparative evaluation of the new GPT-5 family (Nano, Mini, and full) to assess the impact of model scale on diagnostic accuracy and this specific adult-pediatric disparity. MethodsA benchmarking study was conducted using 2,000 multiple-choice questions from the MedQA dataset, equally divided between adult (n=1,000) and pediatric (n=1,000) domains. GPT-5, GPT-5 Mini, and GPT-5 Nano were tested via API with standardized parameters (temperature=0, reasoning effort=minimal, verbosity=low, maxtoken=170). Accuracy was calculated and statistically compared across domains for each model. ResultsA clear dose-response relationship was observed between model size and accuracy. GPT-5 Nano exhibited a significant performance gap, with an accuracy of 71.0% in adult medicine versus 55.4% in pediatrics (a 15.6 percentage point difference, p<0.001). GPT-5 Mini substantially narrowed this gap to 5.7 points (81.5% vs. 75.8%, p=0.001). Critically, the full GPT-5 model eliminated the disparity, achieving comparable accuracy in adult medicine (86.3%) and slightly higher accuracy in pediatrics (88.5%) (p=0.138). Performance gains from scaling up were disproportionately larger for the pediatric domain. ConclusionThe GPT-5 family marks a substantial advancement in medical AI. The full-size model not only achieves high diagnostic accuracy but, crucially, overcomes the previously documented performance limitations in pediatrics. This demonstrates that sufficient model scale is vital for mastering the nuances of specialized clinical domains. These findings support a tiered implementation strategy based on task criticality and underscore the need for continued validation in real-world clinical settings.

Background: Electrolyte replacement is ubiquitous in the acute care setting, but its familiarity cannot belie that even small dosing errors with potassium can cause lethal cardiac arrhythmias. Recently, MedAgentBench offered a benchmark for agentic artificial intelligence (AI) including the ability to correctly dose potassium based on a single rule; however, this does not adequately reflect the clinical complexity or safety concerns of an agent that has been used as the lethal injection. The purpose of this analysis was to a probe leaderboard large language model (LLM) capabilities to follow basic dosing rules to safely replace potassium in a series of clinician-annotated cases. Methods: Using a clinician panel, we developed a series of dosing principles and 20 clinical cases reflective of the complexity of potassium replacement. External clinicians were surveyed to assess practice variability and agreement to clinician panel answers. We tested GPT-5-chat with each case in triplicate, with and without the clinician curated dosing principles, and prompted the model to answer six questions involving potassium goals, dosing, route, lab frequency, concurrent interventions, and the model's perceived level of confidence for the output and complexity of the case. The primary outcome was the rate of appropriate recommendations in comparison to clinician answers. Results: A total of 54 clinicians reviewed the 20 hypokalemia cases and hypokalemia dosing guideline. Clinicians expressed "highly agree" or "somewhat agree" for 66.8% of the cases evaluated when asked if they agree with the guideline-recommended management. When given the potassium dosing guideline, total errors dropped from 165 to 104, and average accuracy improved from 45% to 65% with GPT-5-Chat. GPT-5-Chat conveyed a high level of confidence for 100% of responses, while labeling 80% and 76% of cases as highly complex with and without the criteria, respectively. Potential harm scores were considerable in both groups, however, a notable reduction in severity scores occurred with the dosing guidance document. Recommendations on concurrent interventions and dosing had the highest rate of errors in both groups. Conclusions: Benchmarks must appropriately reflect clinical complexity to be considered valuable for the deployment of agentic artificial intelligence tools in the healthcare domain. GPT-5-Chat assessment on a comprehensive medication management task for potassium replacement showed improvement with dosing guidance, yet unfit benchmarking performance.

Drug repurposing presents a potential solution for finding new therapies for rare and orphan diseases. The limited number of patients affected by rare diseases, combined with scarce research and the financial burden of clinical trials, creates a significant barrier to developing new drugs. Drug repurposing utilizes the known safety profile and effectiveness of existing medications to fast-track the development of life-saving therapies. Recently drug repurposing has focused on utilizing biomedical knowledge graphs to uncover hidden connections between diseases and drugs, revealing promising candidates for repurposing. Because most knowledge graphs in biomedical domain are made by text-mining scientific literature we decided to compare the amount of knowledge contained in open access and controlled (subscription only) access literature. Elsevier and Every Cure make logical partners and allowed the project to use Elseviers ability to access both controlled and open access publications and its proprietary Elsevier AI technology to construct the knowledge graph. Notwithstanding the fact that more than 50% of relationships in drug repurposing for rare diseases can be found in open access content, 45% of relationships remain only in controlled access. We argue that this is due to the large number of edges supported by single reference in the entire biomedical knowledge graph and does not reflect an intrinsic difference between open and controlled access.

PurposeInformation from electronic health records (EHRs) may be incorporated into computable phenotype algorithms in efforts to overcome inaccuracies of algorithms based on administrative claims data alone. However, such efforts can be resource-intensive and unsuccessful. Assessing the feasibility of computable phenotyping for a health outcome of interest (HOI) before proceeding is therefore recommended. MethodsWe developed a systematic fitness-for-purpose (FFP) assessment process to implement concepts outlined in a previously described general framework for computable phenotyping incorporating EHR data. Our process includes verifying the HOI is well-defined, reviewing clinical information about the HOI, identifying existing algorithms and their performance, evaluating HOI clinical and data complexity, and determining an overall FFP conclusion and recommendation. We applied this process to ten HOIs lacking high-performing claims-based algorithms, selecting HOIs of public health importance that varied in clinical and data complexity, including neutropenia, pericardial effusion and drug-induced liver injury. ResultsHOIs assessed as having moderate (vs. easy) overall difficulty had characteristics such as the need for natural language processing, integration of multiple laboratory test results, or longitudinal EHR data. HOIs assessed as having high difficulty required using data from multiple EHR sources, ruling out many other potential causes, or relying on low-sensitivity diagnostic tests. Input from experts in EHR data and clinical care was crucial. ConclusionEHR data have potential to enhance accuracy of defining certain HOIs for research and surveillance compared to administrative claims data. The process and tools we created will support others in assessing FFP of HOIs for computable phenotyping. Five key pointsO_LIIncorporating electronic health record (EHR) data into computable phenotypes could improve accurate identification of health outcomes of interest (HOIs), but such work can be resource intensive. C_LIO_LIWe developed a systematic fitness-for-purpose (FFP) process and tools to assess the feasibility of computable phenotyping for HOIs. C_LIO_LISteps include identifying existing algorithms and their performance, ensuring the HOI is well-defined, evaluating clinical and data complexity, and determining a feasibility recommendation. C_LIO_LIDifficulty increased with a need for natural language processing, multiple laboratory tests, longitudinal EHR data, multiple EHR sources or ruling out other potential causes. C_LIO_LIInput from EHR data and clinical care experts was crucial to the FFP assessment process. C_LI Plain Language Summary (PLS)Attempts to identify diseases and health conditions by applying computer programs to information easily gleaned from insurance claims of tens of thousands of patients (such as FDAs ongoing safety monitoring of approved drugs or medical products) are often unsuccessful because the data lack nuance. Incorporating information from electronic health records (EHR) and patient chart notes may improve accurate identification of health outcomes. Because this can be resource-intensive, we designed a process and tools to assess the feasibility of including EHR data in computer algorithms to identify health outcomes. Steps included identifying existing algorithms and their performance, building familiarity with the outcome and making sure it is well-defined, evaluating clinical and data complexity, and determining a conclusion about feasibility. We applied our process to ten health outcomes of public health importance. Health outcomes were considered moderately difficult for computerized algorithms if they required natural language processing, integration of multiple laboratory tests, or EHR data from multiple timepoints. Health outcomes having high difficulty required using multiple EHR data types, ruling out many alternative causes of the HOI (other than medications), or relying on diagnostic tests of low accuracy. Input from EHR data and clinical care experts was crucial for the assessment process.

BackgroundThe application of large language models (LLMs) in pediatric medicine requires rigorous performance evaluation prior to clinical implementation. ObjectiveTo evaluate the accuracy of the Kimi K2 model in analyzing pediatric clinical cases using a zero-shot approach. Methods: 2,249 multiple-choice questions from pediatric clinical cases, ranging in age from 1 day to 16 years, extracted from the MedQA dataset were analyzed. The model was tested via API with standardized parameters, temperature set to zero, and zero-shot prompts. Accuracy was calculated by comparing the responses with the datasets ground truth. ResultsKimi K2 achieved an overall accuracy of 78.39%, corresponding to 1,763 correct answers out of 2,249 total, with 100% of responses in the required format. Conclusions: The model demonstrates competitive performance for medical education and diagnostic support, while still having limitations that require human clinical supervision.

BackgroundBiomedical Large Language Models (LLMs) combined with prompt engineering offer domain-specific reasoning, yet their application to individual-level causality assessment remains unexplored. This study evaluated five combinations of biomedical LLMs, prompting strategies, and causality algorithms by comparing their agreement with two human expert evaluators. Research design and methodsA total of 150 Individual Case Safety Reports (ICSRs) were analyzed: 140 reports from Food and Drug Administration Adverse Event Reporting System (FAERS), and 10 myocarditis/pericarditis ICSRs from Vaccine AERS (VAERS). Assessments were conducted using the Naranjo and WHO-UMC algorithms. Biomedical LLMs tested included TinyLlama 1.1B, Medicine LLaMA-3 8B, and MedLLaMA v20, combined with Chain-of-Thought (CoT) or Decomposition prompting. Agreement was measured using Gwets Agreement Coefficient 1 (AC1) and percentage agreement, alongside performance metrics and qualitative error analysis. ResultsThe Medicine LLaMA-3 8B-Naranjo-CoT combination achieved the highest agreement with human assessors for the final classification of causality (64%). Biomedical LLMs demonstrated low inter-rater agreement on critical items of causality assessment such as identification of listed AE, temporal plausibility, alternative causes, and objective evidence of AEs. Frequent model failures included irrelevant responses. ConclusionsBiomedical LLMs showed improved performance over general purpose models previously tested but remain suboptimal for reliable causality assessment of ICSRs.

Background: Extrauterine growth restriction (EUGR) is a common and clinically significant complication among preterm infants, contributing to adverse neurodevelopmental and metabolic outcomes. Early and individualized risk prediction remains challenging. This study aimed to develop and validate an interpretable machine learning model for early prediction of EUGR using routinely available clinical variables, and to implement a user-friendly web-based calculator for clinical use. Methods: We retrospectively analyzed 1,431 preterm infants admitted within 24 hours after birth to our hospital between May 2020 and March 2025. Infants from the Yangpu campus (n=863) formed the training set, and those from the Huangpu campus (n=568) formed the validation set. Early clinical variables available within 48-72 hours were screened using the Boruta algorithm. Logistic regression, XGBoost, random forest, decision tree, and support vector machine models were developed and compared. Model performance was evaluated using area under the curve (AUC), accuracy, sensitivity, specificity, F1 score, and Brier score. SHapley Additive exPlanations (SHAP) were applied to assess global and individual feature contributions, nonlinear effects, and interactions. A web-based calculator was constructed based on the optimal model. Results: Nine variables were identified as important predictors: birth weight, small for gestational age status, gestational age, breastfeeding, multiple gestation, neonatal respiratory distress syndrome, patent ductus arteriosus, maternal hypertension, and maternal group B Streptococcus infection. Among the five models, XGBoost achieved the best performance in the validation set (AUC 0.922, accuracy 0.849, Brier score 0.108). SHAP analysis showed that low birth weight, small for gestational age, maternal group B Streptococcus infection, and patent ductus arteriosus were major risk factors, while breastfeeding was protective. Notable nonlinear and interactive effects were observed, particularly between birth weight and gestational age and between breastfeeding and patent ductus arteriosus. The web-based calculator provides real-time individualized risk estimation and visualized interpretation. Conclusions: An interpretable XGBoost-based model and web calculator were successfully developed and validated for early prediction of EUGR in preterm infants. This tool may support clinicians in identifying high-risk infants and guiding individualized nutritional and clinical management.

ObjectiveTo develop and evaluate a personalized literature review system that efficiently processes and summarizes biomedical literature to provide timely, relevant insights for researchers. MethodsThe system integrates ontology-aware keyword extraction (MeSH/ACM constrained TF-IDF from CV/Research Statement), citation-informed retrieval (PubMed and NIH iCite API), and dual-model large language model (LLM) summarization (Google Gemini 2.0 flash, OpenAI GPT-4o-mini). These LLMs leverage advanced Transformer architectures, building on foundations such as BERT, BART, and BioBERT. A two-stage ranking algorithm combines Relative Citation Ratio (RCR) with cosine similarity. Summary quality was evaluated using ROGUE-1/2/L and BERTScore. The system is deployed as a Streamlit web application. ResultsAcross 20 biomedical queries, the system demonstrated strong average performance (BERT-F1{approx} 0.86), with cosine similarity strongly correlating with summary quality. Human evaluation involving 10 users yielded average scores above 4.5/5 across summary fidelity and keyword relevance. ConclusionHybrid ranking and ensemble LLM summarization significantly accelerate scientific sense-making. These findings suggest broad applicability to various domains beyond biomedicine.