Ascertainment of multiple sclerosis (MS) progression is important for informing clinical care decisions and supporting biomedical research. However, the details to infer a patient's MS progression status are locked within clinical notes. In this feasibility study, we assessed the feasibility of developing and validating a large language model (LLM)-based EDSS and FS classifier for ascertaining MS progression from clinical notes.

Multi-omics data are instrumental in obtaining a comprehensive picture of complex biological systems. This is particularly useful for women's health conditions such as endometriosis, which has been historically understudied despite having a high prevalence (around 10% of women of reproductive age). Subsequently, endometriosis has limited genetic characterization: current genome-wide association studies explain only 11% of its 47% total estimated heritability, underscoring the need for integrative approaches. Graph representations provide an intuitive and meaningful way to harmonize biological data, using nodes to represent biological concepts (e.g., genes, single nucleotide polymorphisms, proteins, and phenotypes) and edges to represent their relationships. We present DRIVE-KG (Disease Risk Inference and Variant Exploration Knowledge Graph), which uses a heterogeneous graph representation to integrate data from diverse multi-omics datasets. We trained two distinct models using DRIVE-KG: a link prediction model to suggest associations between SNPs and two pilot phenotypes (endometriosis and obesity), and a graph convolutional network (GCN) for patient-level classification of endometriosis/adenomyosis as a combined phenotype. We conducted patient-level classification using data from 1,441 Penn Medicine BioBank participants with gold standard chart-reviewed endometriosis/adenomyosis status. The link prediction model uncovered 66 high-confidence (model score ≥ 0.95) candidate SNP-endometriosis associations, representing largely distinct genetic signals (R2 < 0.1). These variants were enriched for obesity/body mass index traits (24.2%), lipid metabolism (6%), and depressive disorders (4.5%), showing agreement with emerging hypotheses about endometriosis etiology. In contrast, of the high-confidence, candidate SNP-obesity associations that could be evaluated using LDlink, 38.22% were in high linkage disequilibrium (R2 ≥ 0.8) with known obesity or comorbidity associations. The GCN to classify patient endometriosis/adenomyosis status had an F1 score of 0.752 compared to 0.698 for a genetic risk score. Despite this moderate improvement, we found that the GCN learned meaningful stratification of underlying adenomyosis signal and severe endometriosis grades. Together, these results demonstrate that heterogeneous integration of multi-omics data is valuable for diverse downstream tasks-including discovery and clinical prediction-particularly for understudied diseases where traditional genomic approaches are insufficient.

Persistent demographic disparities have been identified in the treatment of patients seeking care in the emergency department (ED). These may be driven in part by subconscious biases, which providers themselves may struggle to identify. To better understand the operation of these biases, we performed a retrospective cross-sectional analysis using electronic health records describing 339,400 visits to the ED of a single US pediatric medical center between 2019-2024. Odds ratios were calculated using propensity-score matching. Analyses were adjusted for confounding variables, including chief complaint, insurance type, socioeconomic deprivation, and patient comorbidities. We also trained a machine learning [ML] model on this dataset to identify predictors of admission. We found significant demographic disparities in admission (Non-Hispanic Black [NHB] relative to Non-Hispanic White [NHW]: OR 0.77, 95% CI 0.73-0.81; Hispanic relative to NHW: OR 0.80, 95% CI 0.76-0.83). We also identified disparities in individual decisions taken during the ED stay. For example, NHB patients were significantly less likely than NHW patients to be assigned an 'emergent' triage acuity score of (OR 0.70, 95% CI 0.67-0.72), but emergent NHB patients were also significantly less likely to be admitted than NHW patients with the same triage acuity (OR 0.86, 95% CI 0.80-0.93). Demographic disparities were particularly acute wherever patients had normal vital signs, public insurance, moderate socio-economic deprivation, or a home address distant from the hospital. An ML model assigned higher importance to triage score for NHB than NHW patients when predicting admission, reflecting these disparities in assignment. We conclude that many visit characteristics, clinical and otherwise, may influence the operation of subconscious biases and affect ML-driven decision support tools.

Pinpointing causal genes at genome-wide association study (GWAS) loci remains a major bottleneck. Existing literature-mining approaches are often limited in accuracy and scalability. We show that large language models (LLMs) can accurately prioritize likely causal genes at GWAS loci. We systematically evaluated several widely available general-purpose LLMs against benchmark datasets of high-confidence causal genes, including a unique set from 23 unpublished GWAS. Our results demonstrate that LLMs outperform or match current state-of-the-art methods and, crucially, exhibit robust performance on novel loci not previously linked to traits, underscoring their generalizability. Moreover, when integrated with existing methods, LLMs substantially enhance overall performance. This work establishes LLMs as an accurate, scalable, and broadly generalizable approach to accelerate causal gene identification in complex traits.

We propose a novel strategy for provenance tracing in random walk-based network diffusion algorithms, a problem that has been surprisingly overlooked in spite of the widespread use of diffusion algorithms in biological applications. Our path-based approach enables ranking paths by the magnitude of their contribution to each node's score, offering insight into how information propagates through a network. Building on this capability, we introduce two quantitative measures: (i) path-based effective diffusion, which evaluates how well a diffusion algorithm leverages the full topology of a network, and (ii) diffusion betweenness, which quantifies a node's importance in propagating scores. We applied our framework to SARSCoV- 2 protein interactors and human PPI networks. Provenance tracing of the Regularized Laplacian and Random Walk with Restart algorithms revealed that a substantial amount of a node's score is contributed via multi-edge paths, demonstrating that diffusion algorithms exploit the non-local structure of the network. Analysis of diffusion betweenness identified proteins playing a critical role in score propagation; proteins with high diffusion betweenness are enriched with essential human genes and interactors of other viruses, supporting the biological interpretability of the metric. Finally, in a signaling network composed of causal interactions between human proteins, the top contributing paths showed strong overlap with COVID-19-related pathways. These results suggest that our path-based framework offers valuable insight into diffusion algorithms and can serve as a powerful tool for interpreting diffusion scores in a biologically meaningful context, complementing existing module- or node-centric approaches in systems biology. The code is publicly available at https://github.com/n-tasnina/provenance-tracing.git under the GNU General Public License v3.0.

When evaluations aren't trustworthy, entire research programs can chase mirages. Objective benchmarks and independent assessment have repeatedly catalyzed progress across computational biology, from protein structure prediction to variant interpretation and single‑cell analysis. This workshop gathers leaders of community challenges and benchmarking infrastructures together with domain experts to provide a contemporary view of how to design trustworthy evaluations, why blind prediction matters, and where standards, infrastructure, and policy must evolve to meet the demands of AI‑driven biology. We summarize the motivation and scope of the workshop; provide background on methodological and infrastructural advances that enable rigorous benchmarking; highlight invited speakers' contributions; and outline anticipated outcomes and community calls‑to‑action.

The ongoing need for effective epidemic modeling has driven advancements in capturing the complex dynamics of infectious diseases. Traditional models, such as Susceptible-Infected- Recovered, and graph-based approaches often fail to account for higher-order interactions and the nuanced structure pattern inherent in human contact networks. Higher-order interactions, such as those in schools, workplaces, or public transit, involve simultaneous contact among more than two individuals. This study introduces a novel Human Contact-Tracing Hypergraph Neural Network framework tailored for epidemic modeling called EpiDHGNN, leveraging the capabilities of hypergraphs to model intricate, higher-order relationships from both location and individual level. Both real-world and synthetic epidemic data are used to train and evaluate the model. Results demonstrate that EpiDHGNN consistently outperforms baseline models across various epidemic modeling tasks, such as source detection and forecast, by approximately 12.1% through effectively capturing the higher-order interactions and preserving the complex structure of human interactions. This work underscores the potential of representing human contact data as hypergraphs and employing hypergraphbased methods to improve epidemic modeling, providing reliable insights for public health decision-making.

Alzheimer’s Disease and Related Dementias (ADRD) afflict almost 7 million people in the USA alone. The majority of research in ADRD is conducted using post-mortem samples of brain tissue or carefully recruited clinical trial patients. While these resources are excellent, they suffer from lack of sex/gender, and racial/ethnic inclusiveness. Electronic Health Records (EHR) data has the potential to bridge this gap by including real-world ADRD patients treated during routine clinical care. In this study, we utilize EHR data from a cohort of 70,420 ADRD patients diagnosed and treated at Penn Medicine. Our goal is to uncover important risk features leading to three types of Neuro-Degenerative Disorders (NDD), including Alzheimer’s Disease (AD), Parkinson’s Disease (PD) and Other Dementias (OD). We employ a variety of Machine Learning (ML) Methods, including uni-variate and multivariate ML approaches and compare accuracies across the ML methods. We also investigate the types of features identified by each method, the overlapping features and the unique features to highlight important advantages and disadvantages of each approach specific for certain NDD types. Our study is important for those interested in studying ADRD and NDD in EHRs as it highlights the strengths and limitations of popular approaches employed in the ML community. We found that the uni-variate approach was able to uncover features that were important and rare for specific types of NDD (AD, PD, OD), which is important from a clinical perspective. Features that were found across all methods represent features that are the most robust.

Adverse drug responses (ADRs) result in over 7,000 deaths annually. Pharmacogenomic studies have shown that many ADRs are partially attributable to genetics. However, emerging data suggest that epigenetic mechanisms, such as DNA methylation (DNAm) also contribute to this variance. Understanding the impact of DNA methylation on drug response may minimize ADRs and improve the personalization of drug regimens. In this work, we identify DNA methylation sites that likely impact drug response phenotypes for anticoagulant and cardiometabolic drugs. We use instrumental variable analysis to integrate genome-wide association study (GWAS) summary statistics derived from electronic health records (EHRs) within the U.K. Biobank (UKBB) with methylation quantitative trait loci (mQTL) data from the Genetics of DNA Methylation Consortium (GoDMC). This approach allows us to achieve a robust sample size using the largest publicly available pharmacogenomic GWAS. For warfarin, we find 71 DNAm sites. Of those, 8 are near the gene VKORC1 and 48 are on chromosome 6 near the human leukocyte antigen (HLA) gene family. We also find 2 warfarin DNAm sites near the genes CYP2C9 and CYP2C19. For statins, we identify 17 DNAm sites. Eight are near the APOB gene, which encodes a carrier protein for low-density lipoprotein cholesterol (LDL-C). We find no novel significant epigenetic results for metformin.

Alzheimer's disease (AD) is a polygenic disorder with a prolonged prodromal phase, complicating early diagnosis. Recent research indicates that increased astrocyte reactivity is associated with a higher risk of pathogenic tau accumulation, particularly in amyloid-positive individuals. However, few clinical tools are available to predict which individuals are likely to exhibit elevated astrocyte activation and, consequently, be susceptible to hyperphosphorylated tau-induced neurodegeneration. Polygenic risk scores (PRS) aggregate the effects of multiple genetic loci to provide a single, continuous metric representing an individual's genetic risk for a specific phenotype. We hypothesized that an astrocyte activation PRS could aid in the early detection of faster clinical decline. Therefore, we constructed an astrocyte activation PRS and assessed its predictive value for cognitive decline and AD biomarkers (i.e., cerebrospinal fluid [CSF] levels of Aβ1-42, total tau, and p-tau181) in a cohort of 791 elderly individuals. The astrocyte activation PRS showed significant main effects on cross-sectional memory (β = -0.07, p = 0.03) and longitudinal executive function (β = -0.01, p = 0.03). Additionally, the PRS interacted with amyloid positivity (p.intx = 0.02), whereby indicating that amyloid burden modifies the association between the PRS and annual rate of language decline. Furthermore, the PRS was negatively associated with CSF Aβ1-42 levels (β = -3.4, p = 0.07) and interacted with amyloid status, such that amyloid burden modifies the association between the PRS and CSF phosphorylated tau levels (p.intx = 0.08). These findings suggest that an astrocyte activation PRS could be a valuable tool for early disease risk prediction, potentially enabling intervention during the interval between pathogenic amyloid and tau accumulation.