BioData Mining

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.

In the management of atrial fibrillation, the most frequently prescribed oral anticoagulant is apixaban, given at a fixed dose of 5mg BID. Apixaban is predominantly metabolized by cytochrome P4503A4 (CYP3A4) and is also a substrate for the drug efflux transporter P-glycoprotein (P-gp). In nearly 300,000 Medicare patients with AF receiving apixaban, we previously showed that concomitant therapy with drugs that inhibit both CYP3A4 and P-gp, specifically amiodarone or diltiazem, significantly increased serious bleeding that caused hospitalization and/or death. We hypothesized that this adverse effect was mediated by an increase in apixaban plasma concentrations caused by concomitant therapy that reduced drug elimination. Utilizing left-over samples obtained from clinically indicated blood draws that would typically be discarded, the Vanderbilt University Medical Center biobank BioVU contains >353,000 samples linked to de-identified electronic medical records (EMRs), with both DNA and plasma harvested. Of 35 samples drawn from patients taking apixaban 5mg BID, 5 were identified to be drawn from patients concomitantly taking drugs inhibiting both CYP3A4 and P-gp. Using a chromogenic anti-Xa assay, we found that plasma concentrations of apixaban were significantly higher (347{+/-}64 ng/mL; mean{+/-}SEM) for patients receiving concomitant CYP3A4/P-gp-inhibiting drugs compared to those not treated with these drugs (166{+/-}67 ng/mL; P=0.025, Mann Whitney). There were no differences between the 2 patient groups with respect to age, weight, or serum creatinine. The results of this pilot study provide preliminary data to support our hypothesis, and they demonstrate the practicality of obtaining pharmacokinetic data from a large cohort of plasma samples linked to deidentified EMRs. This approach could be used to define the role of apixaban levels in high-risk clinical scenarios and to better understand the relationship between drug levels and bleeding risk.

BackgroundAccurate single-nucleotide polymorphism (SNP) annotation is central to genomic research yet widely used tools and gene models often yield divergent results. Prior studies have shown such discrepancies in small datasets, but the extent of genome-wide variation and its impact on downstream pathway analysis remain unclear. ResultsWe conducted a comprehensive comparison of three commonly used SNP annotation tools, ANNOVAR, SnpEff, and VEP, using both Ensembl and RefSeq gene models to evaluate more than 40 million SNPs from the Haplotype Reference Consortium. At the protein level, annotation output differed significantly across tools and gene models (p-adj < 0.001), with discrepancies present in both genic and intergenic regions. RefSeq produced broader annotation coverage, particularly for intergenic SNPs, while Ensembl showed greater internal consistency. SnpEff provided the most complete coverage overall, whereas no single tool or model configuration achieved full annotation recovery of the union reference. Integration across tools and models maximized coverage and reduced annotation loss. In a case study of 204 colorectal cancer-associated SNPs from the FIGI GWAS, pathway enrichment results varied depending on annotation strategy. The fully integrated approach identified all four significant pathways, whereas several single-tool or single-model strategies missed one or more. ConclusionSNP annotation outcomes are influenced by both the tool and gene model used, and relying on a single approach may result in incomplete coverage. A multi-tool, multi-model strategy provides the most comprehensive annotation and preserves enriched pathways, supporting more robust and reproducible genomic interpretation.

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.

ObjectiveSNP heritability estimates vary substantially across estimation strategies, yet the downstream consequences for polygenic risk score (PRS) construction remain poorly characterised. We systematically benchmarked heritability estimation configurations and assessed their propagation into downstream PRS performance. MethodsWe benchmarked 86 heritability-estimation configurations spanning six tool families (GEMMA, GCTA, LDAK, DPR, LDSC, SumHer) and ten method groups across 10 UK Biobank phenotypes, yielding 844 configuration-level estimates. Each estimate was propagated into GCTA-SBLUP and LDpred2-lassosum2 PRS frameworks and evaluated across five cross-validation folds using null, PRS-only, and full models. Eleven binary analytical contrasts were tested using Mann-Whitney U tests to identify drivers of heritability variability. ResultsHeritability ranged from -0.862 to 2.735 (mean = 0.134, SD = 0.284), with 133 of 844 estimates (15.8%) negative and concentrated in unconstrained estimation regimes. Ten of eleven analytical contrasts significantly affected heritability magnitude, with algorithm choice and GRM standardisation showing the largest effects. Despite this upstream variability, downstream PRS test performance was only weakly coupled to heritability magnitude: pooled Pearson correlations between h2 and test AUC were r = -0.023 for GCTA-SBLUP and r = +0.014 for LDpred2-lassosum2 (both non-significant). ConclusionSNP heritability is best interpreted as a configuration-sensitive modelling parameter rather than a universally stable scalar input. Heritability estimates should always be reported alongside their full estimation specification, and downstream PRS performance is comparatively robust to moderate variation in the heritability input. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=80 SRC="FIGDIR/small/716079v1_ufig1.gif" ALT="Figure 1"> View larger version (27K): org.highwire.dtl.DTLVardef@112929borg.highwire.dtl.DTLVardef@573c36org.highwire.dtl.DTLVardef@132170borg.highwire.dtl.DTLVardef@1871363_HPS_FORMAT_FIGEXP M_FIG C_FIG

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.

BackgroundDespite extensive cataloging of carcinogenic exposures by the International Agency for Research on Cancer (IARC) and pharmacogenomic variation by resources such as PharmVar and CPIC, few platforms unify exposure, metabolic activation and detoxification, DNA damage, and genetic annotation within a single interactive visualization framework. This gap limits systematic evaluation of gene-environment interactions in cancer risk assessment. MethodsWe developed the Carcino-Genomic Knowledge Graph, ExposoGraph, an interactive knowledge-graph platform for carcinogen metabolism and DNA damage pathways. The reference graph integrates curated data and annotations from IARC, KEGG, PharmVar, CPIC, CTD, and supporting literature/resources. The current reference graph contains 96 nodes across 5 entity types (Carcinogens, Enzymes, Metabolites, DNA Adducts, and Pathways) and 102 edges across 6 relationship types (activates, detoxifies, transports, forms adduct, repairs, and pathway). ResultsThe first-generation reference graph captures metabolic activation and detoxification pathways for 9 carcinogen classes spanning 15 index carcinogens. It represents 36 enzymes across Phase I activation (n=14), Phase II conjugation and detoxification (n=14), Phase III transport (n=3), and DNA repair (n=5). Interactive exploration supports carcinogen-class filtering, node- and edge-type filtering, metadata-based search, and detailed hover/detail views with provenance and pharmacogenomic annotations. The androgen branch highlights cross-pathway connectivity by linking androgen metabolism to estrogen quinone formation and DNA adduct generation through CYP19A1-mediated aromatization and downstream catechol estrogen chemistry. In the optional androgen-focused extension, additional receptor, tissue, and variant context further connects this branch to androgen receptor signaling and genotype-specific annotations. ConclusionsExposoGraph provides a first-generation integrated, interactive framework linking carcinogenic exposures to metabolic fates and genetic modulators. The platform supports hypothesis generation for gene-environment interaction studies and may inform future individualized risk modeling, while remaining a research-use framework rather than a clinically validated risk-assessment tool.

ImportanceGuideline-concordant care for young children with attention-deficit/hyperactivity disorder (ADHD) includes recommending parent training in behavior management (PTBM) as first-line treatment. However, assessing guideline adherence through manual chart review is time-consuming and costly, limiting scalable and timely quality-of-care measurement. ObjectiveTo evaluate the accuracy and explainability of large language models (LLMs) in identifying PTBM recommendations in pediatric electronic health record (EHR) notes as a scalable alternative to manual chart review. Design, Setting, and ParticipantsThis retrospective cohort study was conducted in a community-based pediatric healthcare network in California consisting of 27 primary care clinics. The study cohort included children aged 4-6 years with [&ge;] 2 primary care visits between 2020-2024 and ICD-10 diagnoses of ADHD or ADHD symptoms (n=542 patients). Clinical notes from the first ADHD-related visit were included. A stratified subset of 122 notes, including all cases with model disagreement, was manually annotated to assess model performance in identifying PTBM recommendations and rank model explanations. ExposuresAssessment and plan sections of clinical notes were analyzed using three generative large language models (Claude-3.5, GPT-4o, and LLaMA-3.3-70B) to identify the presence of PTBM recommendations and generate explanatory rationales and documentation evidence. Main Outcomes and MeasuresModel performance in identifying PTBM recommendations (measured by sensitivity, positive predictive value (PPV), and F1-score) and qualitative explainability ratings of model-generated rationales (based on the QUEST framework). ResultsAll three models demonstrated high performance compared to expert chart review. Claude-3.5 showed balanced performance (sensitivity=0.89, PPV=0.95, and F1-score=0.92) and ranked highest in explainability. LLaMA3.3-70B achieved sensitivity=0.91, PPV=0.89, and F1-score=0.90, ranking second for explainability. GPT-4o had the highest PPV [0.97] but lowest sensitivity [0.82], with an F1-score of 0.89 and the lowest explainability ranking. Based on classifications from the best-performing model, Claude-3.5, 26.4% (143/542) of patients had documented PTBM recommendations at their first ADHD-related visit. Conclusions and RelevanceLLMs can accurately extract guideline-concordant clinician recommendations for non-pharmacological ADHD treatment from unstructured clinical notes while providing clear explanations and supporting evidence. Evaluating model explainability as part of LLM implementation for medical chart review tasks can promote transparent and scalable solutions for quality-of-care measurement.

Machine learning-based annotation methods are increasingly used to assess the pathogenicity of genetic variants, but their performance at prioritizing variants for gene-level association testing remains poorly characterized. Here, we systematically benchmark five annotation methods -- CADD v1.6, CADD v1.7, AlphaMissense, ESM-1b, and GPN-MSA -- using four primary gene-based tests and six annotation-level aggregation tests across 14 quantitative traits measured in up to 350,377 UK Biobank participants. Using a novel framework based on Wasserstein dis-tances, we quantify how annotation choice affects test calibration and power. Tests using CADD annotations achieve the highest signal separation, while tests using AlphaMissense annotations exhibit systematically lower calibration. All combinations of methods produced significant re-sults that were enriched (1.8-5.8-fold) for loss-of-function intolerant genes, though tests using GPN-MSA annotations displayed the highest such enrichment. Replication across symmetric phenotypes and loss-of-function burden tests was generally similar across methods. Our anal-ysis provides practical guidance for annotation method selection in rare variant studies and establishes a distributional framework for calibration assessment.

The goal of this work was to leverage a large corpus of text based psychotherapy data to create novel machine learning algorithms that can identify suicide risk in asynchronous text therapy. Advances in the field of natural language processing and machine learning have allowed us to include novel data sources as well as use encoding models that can represent context. Our models utilize advanced natural language processing techniques, including fine-tuned transformer models like RoBERTa, to classify risk. Subsequent model versions incorporated non-text data such as demographic features and census-derived social determinants of health to improve equitable and culturally responsive risk assessment, as well as multiclass models that can identify tiered levels of risk. All new models demonstrated significant improvements over our previous model. Our final version, a multiclass model, provides a tiered system that classifies risk as "no risk," "moderate," or "severe" (weighted F1 of 0.85). This tiered approach enhances clinical utility by allowing providers to quickly prioritize the most urgent cases, ensuring a more accurate and timely intervention for clients in need.

BackgroundMixed-format lifestyle questionnaires contain both structured variables and free-text responses, but it remains unclear whether language-derived variables provide incremental predictive value beyond structured data, and under which representational condition. It was investigated whether variables derived from patient-reported free text improve ALS-versus-control classification beyond structured questionnaire data, and whether their value depends on how temporal information is represented. MethodsA leakage-free machine-learning pipeline was developed to classify ALS versus controls from questionnaire-derived data, including a schema-guided LLM-based text-to-table extraction and a compact longitudinal encoding strategy. Three feature configurations were compared: Pool1, containing structured baseline variables only; Pool2, adding compact summaries derived from first-time-point (T1) free-text responses; and Pool3, further incorporating compact descriptors of change between T1 and T2. Logistic Regression, linear Support Vector Classification, and Random Forest were evaluated using repeated stratified holdout (10 seeds) and repeated stratified 5-fold cross-validation. Final ablation analyses were performed to isolate the contribution of the compact text block and the compact temporal block. ResultsAfter leakage correction, performance estimates became more conservative, indicating that previous results had been optimistic. In the final configuration, Pool3 achieved the best performance, with Random Forest reaching a holdout accuracy of 0.673, F1-weighted score of 0.666, and Matthews correlation coefficient of 0.323; cross-validated F1-weighted score and Matthews correlation coefficient were 0.654 and 0.312, respectively. Pool2 did not show a robust improvement over Pool1. Ablation analysis showed that removing the compact temporal block markedly reduced Pool3 performance, whereas removing the compact text block had little overall effect. These findings indicate that the primary value of language-based processing in small clinical cohorts lies not in static feature enrichment, but in enabling compact representations of longitudinal change. ConclusionsIn this setting, the main predictive gain did not arise from static text-derived variables alone, but from representing questionnaire information as compact longitudinal change descriptors. These findings suggest that, in small clinical cohorts, the value of language-based processing may lie more in summarizing trajectories than in expanding static feature spaces.

Accurate classification of SARS-CoV-2 genomic variants is essential for effective genomic surveillance, yet it is challenged by extreme class imbalance, limited representation of rare variants, and distribution shifts in real-world sequencing data. In this study, we employed hybrid RF-SVM framework designed for robust detection of rare SARS-CoV-2 variants. It integrates a random forest and a polynomial-kernel based support vector machine to enhance sensitivity to minority classes while maintaining overall predictive stability. We systematically compared classical machine learning models, deep learning approaches, and hybrid strategies under both standard and distribution-shifted evaluation settings. Our results show that classical models using TF-IDF-based k-mer features outperform deep learning methods on macro-averaged performance metrics. The Random Forest classifier using TF-IDF Feature achieved the best overall performance, with a macro-averaged F1-score of 0.8894 and an accuracy of 96.3%. The model also demonstrated strong generalization ability, as evidenced by stable cross-validation performance (CV accuracy = 0.9637). Hybrid RF-SVM model further improves rare variant detection under severe class imbalance. Calibration analysis indicates reliable probability estimates for common variants, although challenges persist for minority classes. Overall, this study highlights the limitations of deep learning in highly imbalanced genomic settings and demonstrates that carefully designed hybrid machine learning approaches provide an effective and interpretable solution for rare SARS-CoV-2 variant detection.

Background: Large language models (LLMs) are increasingly used in mental health contexts, yet their detection of suicidal ideation is inconsistent, raising patient safety concerns. Objective: To evaluate whether an independent safety monitoring system improves detection of suicide risk compared with native LLM safeguards. Methods: We conducted a cross-sectional evaluation using 224 paired suicide-related clinical vignettes presented in a single-turn format under two conditions (with and without structured clinical information). Native LLM safeguard responses were compared with an independent supervisory safety architecture with asynchronous monitoring. The primary outcome was detection of suicide risk requiring intervention. Results: The supervisory system detected suicide risk in 205 of 224 evaluations (91.5%) versus 41 of 224 (18.3%) for native LLM safeguards. Among 168 discordant evaluations, 166 favored the supervisory system and 2 favored the LLM (matched odds ratio {approx}83.0). Both systems detected risk in 39 evaluations, and neither in 17. Detection was highest in scenarios with explicit suicidal ideation and lower in more ambiguous presentations. Conclusions: Native LLM safeguards frequently failed to detect suicide risk in this structured evaluation. An independent monitoring approach substantially improved detection, supporting the role of external safety systems in high-risk mental health applications of LLMs.

Automating Hierarchical Condition Category (HCC) assignment directly from unstructured electronic health record (EHR) notes remains an important but understudied problem in clinical informatics. We present HCC-Coder, an end to end NLP system that maps narrative documentation to 115 Centers for Medicare & Medicaid Services(CMS) HCC codes in a multi-label setting. On the test dataset, HCC-Coder achieves a macro-F1 of 0.779 and a micro-F1 of 0.756, with a macro-sensitivity of 0.819 and macro-specificity of 0.998. By contrast, Generative Pre-trained Transformer (GPT)-4o achieves highest score of a macro-F1 of 0.735 and a micro-F1 of 0.708 under five-shot prompting. The fine-tuned model demonstrates consistent absolute improvements of 4%-5% in F1-scores over GPT-4o. To address severe label imbalance, we incorporate inverse-frequency weighting and per-label threshold calibration. These findings suggest that domain-adapted transformers provide more balanced and reliable performance than prompt-based large language models for hierarchical clinical coding and risk adjustment.

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).

Polygenic risk score (PRS) tools differ substantially in statistical assumptions, input requirements, and implementation complexity, making direct comparison difficult. We developed a harmonized, implementation-aware benchmarking framework to evaluate 46 PRS tools across seven binary UK Biobank phenotypes and one continuous trait under three model configurations: null, PRS-only, and PRS plus covariates. The framework integrates standardized preprocessing, tool-specific execution, hyperparameter exploration, and unified downstream evaluation using five-fold cross-validation on high-performance computing infrastructure. In addition to predictive performance, we assessed runtime, memory use, input dependencies, and failure modes. A Friedman test across 40 phenotype-fold combinations confirmed significant differences in tool rankings ({chi}2 = 102.29, p = 2.57 x 10-11), with no single method universally optimal. These findings provide a reproducible framework for comparative PRS evaluation and demonstrate that tool performance is shaped not only by statistical methodology but also by phenotype architecture, preprocessing choices, covariate structure, computational demands, software robustness, and practical implementation constraints.

MotivationCell-cell interaction (CCI) underlies several fundamental mechanisms including development, homeostasis and disease progression. CCI are known to be localised to specific subcellular regions, for example, within the cytoplasms of cells. With the emergence of subcellular spatial transcriptomics technologies (sST), there is an opportunity to attribute CCI to subcellular regions. We aimed to deconvolute CCI to subcellular CCI (sCCI) in non-spatial single cell transcriptomics data (i.e. scRNA-seq) datasets using a modified CCI score from CellChat. ResultsBy calculating the sCCI score specific to cytoplasm and nucleus in nine publicly available sST datasets, we identified unique nucleus-nucleus and cytoplasm-cytoplasm sCCI. Then, we deconvolved the communication score to subcellular regions by using a hierarchical classification and regression model which we name as CCIDeconv. We performed leave-one-dataset-out cross-validation across nine datasets over a range of different tissue types from human samples. We observed that training across many different tissue types resulted in robust deconvolution performance in an unseen dataset. As the number of training datasets increased, models trained without spatial features achieved similar performance as models including spatial features. This implied the potential for accurate prediction of sCCI events from even scRNA-seq with large numbers of training datasets. Overall, we offer a method towards attributing CCI events to subcellular regions. This method can allow researchers in dissecting sCCI patterns to gain insights in underlying biology in a range of tissues covering health and disease.

Background: The FDA Adverse Event Reporting System (FAERS) is a critical pillar of post-marketing pharmacovigilance; however, its utility is constrained by data heterogeneity, pervasive reporting redundancies, and inconsistent medical terminology. These structural barriers impede reproducible, large-scale analyses and the implementation of precision drug safety surveillance. Methods: We developed faers, an open-source R package that delivers a standardized framework and an end-to-end workflow for transforming raw FAERS data into analysis-ready formats. The package implements a regulatory-compliant multi-level deduplication strategy, automated MedDRA terminology mapping, and an R S4-based object-oriented system to ensure data integrity, traceability, and efficient management of complex relational structures. It further integrates a full suite of disproportionality signal detection methods, including the Reporting Odds Ratio (ROR), Proportional Reporting Ratio (PRR), Bayesian Confidence Propagation Neural Network (BCPNN), and Empirical Bayes Geometric Mean (EBGM). Performance was benchmarked on large-scale FAERS datasets, and validity was confirmed by replicating published findings on anti-PD-1/PD-L1-associated cardiotoxicity and CAR-T cell therapy outcomes, with additional application to immune-related adverse events (irAEs). Findings: The package demonstrated high computational efficiency and near-linear scalability when processing extensive quarterly FAERS data. Validation analyses of two case studies showed excellent concordance with prior literature. Application to an irAE cohort further identified a statistically significant age-by-sex interaction in risk patterns, demonstrating the tool's ability to uncover nuanced demographic signals that are often missed by conventional approaches. Interpretation: The faers package provides a transparent, scalable, and fully reproducible framework for FAERS-based pharmacovigilance. By automating data cleaning, standardization, and advanced signal detection, it lowers technical barriers for researchers and regulators while promoting high-quality, open pharmacoepidemiological research to strengthen drug safety monitoring.

Purpose: Suicide and self-harm are major public health concerns characterized by substantial clinical and psychosocial heterogeneity. While latent class analysis has been used to identify subgroups of people with suicidal behavior, the extent to which such population-level phenotyping complements explainable artificial intelligence-based classification models remain unclear. Methods: We applied latent class analysis to a cross-sectional, publicly available dataset of 1000 individuals presenting with self-harm and suicide-related behaviors at Colombo South Teaching Hospital, Kalubowila, Sri Lanka. Sociodemographic, psychosocial, and clinical variables were used to identify latent subgroups. Class characteristics and suicide prevalence were examined and compared with variable importance patterns reported in a previously published explainable artificial intelligence (XAI)-based suicide classification study using the same dataset. Results: Four latent classes were identified. Two classes exhibited very high suicide prevalence (91.2% [95% CI: 87.7-93.8] and 99.0% [95% CI: 96.4-99.7]), whereas two classes showed low prevalence (<1%). The two high-prevalence classes differed markedly in lifetime psychiatric hospitalization history, with one class showing a 100% prevalence of prior hospitalization and the other substantially lower hospitalization rates. These patterns partially aligned with, and extended beyond, variable importance findings from the XAI-based model. Conclusion: Latent class analysis identified distinct subgroups with substantially different suicide prevalence and clinical profiles, underscoring the heterogeneity of individuals presenting with self-harm. Comparison with XAI-based suicide classification model findings suggest that unsupervised phenotyping and supervised classification provide complementary perspectives, offering population-level context that may enhance the interpretability of suicide assessment frameworks. Keywords: suicide; self-harm; latent class analysis; explainable artificial intelligence; machine learning

Background: Large language models (LLMs) contain limited professional medical knowledge, as large-scale training on clinical text has not yet been possible due to restricted access. Objectives: To continue pre-training an open-access instruct LLM on de-identified medical notes and evaluate the resulting impact on real-world clinical decision-making tasks and standard benchmarks. Methods: Using 500K de-identified clinical notes from Cedars-Sinai Health System, we fine-tuned a Qwen3-4B Instruct model with supervised learning to generate medical decision-making (MDM) paragraphs from patient presentations, and evaluated it on assigned-diagnosis prediction, in-hospital cardiac-arrest mention detection, and a suite of general and biomedical benchmarks. Results: The fine-tuned model produced MDMs that closely resembled those written by physicians and outperformed the base-instruct model and larger clinically untrained models (Qwen3-32B and Llama-3.1-405B Instruct) on assigned-diagnosis prediction, the task most aligned with its training objective. On the task of detecting in-hospital cardiac arrest mentions, the model initially exhibited mild label collapse, but a brief task-specific fine-tuning stage resolved this issue and allowed it to surpass all competitors. The model also demonstrated global general knowledge retention on biomedical and general-domain evaluation benchmarks compared to the baseline. Conclusion: Supervised full fine-tuning on clinical notes allowed the model to incorporate medical knowledge without sacrificing general-domain abilities, and to transfer this knowledge to unseen biomedical tasks without wholesale loss of general-domain abilities, while revealing collapse-related failure modes that motivate more principled strategies for clinical specialization.