Molecular representation learning is vital for various downstream applications, including the analysis and prediction of molecular properties and side effects. While Graph Neural Networks (GNNs) have been a popular framework for modeling molecular data, they often struggle to capture the full complexity of molecular representations. In this paper, we introduce a novel method called Gode, which accounts for the dual-level structure inherent in molecules. Molecules possess an intrinsic graph structure and simultaneously function as nodes within a broader molecular knowledge graph. Gode integrates individual molecular graph representations with multi-domain biochemical data from knowledge graphs. By pre-training two GNNs on different graph structures and employing contrastive learning, Gode effectively fuses molecular structures with their corresponding knowledge graph substructures. This fusion yields a more robust and informative representation, enhancing molecular property predictions by leveraging both chemical and biological information. When fine-tuned across 11 chemical property tasks, our model significantly outperforms existing benchmarks, achieving an average ROC-AUC improvement of 12.7% for classification tasks and an average RMSE/MAE improvement of 34.4% for regression tasks. Notably, Gode surpasses the current leading model in property prediction, with advancements of 2.2% in classification and 7.2% in regression tasks.

Aofei Chang, Le Huang, Alex James Boyd, Parminder Bhatia, Taha Kass-Hout, Cao Xiao, Fenglong Ma. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Foundational models such as the Segment Anything Model (SAM) are gaining traction in medical imaging segmentation, supporting multiple downstream tasks. However, such models are supervised in nature, still relying on large annotated datasets or prompts supplied by experts. Conventional techniques such as active learning to alleviate such limitations are limited in scope and still necessitate continuous human involvement and complex domain knowledge for label refinement or establishing reward ground truth. To address these challenges, we propose an enhanced Segment Anything Model (SAM) framework that utilizes annotation-efficient prompts generated in a fully unsupervised fashion, while still capturing essential semantic, location, and shape information through contrastive language-image pretraining and visual question answering. We adopt the direct preference optimization technique to design an optimal policy that enables the model to generate high-fidelity segmentations with simple ratings or rankings provided by a virtual annotator simulating the human annotation process. State-of-the-art performance of our framework in tasks such as lung segmentation, breast tumor segmentation, and organ segmentation across various modalities, including X-ray, ultrasound, and abdominal CT, justifies its effectiveness in low-annotation data scenarios.

Large language models (LLMs) have demonstrated significant potential in clinical decision support. Yet LLMs still suffer from hallucinations and lack fine-grained contextual medical knowledge, limiting their high-stake healthcare applications such as clinical diagnosis. Traditional retrieval-augmented generation (RAG) methods attempt to address these limitations but frequently retrieve sparse or irrelevant information, undermining prediction accuracy. We introduce KARE, a novel framework that integrates knowledge graph (KG) community-level retrieval with LLM reasoning to enhance healthcare predictions. KARE constructs a comprehensive multi-source KG by integrating biomedical databases, clinical literature, and LLM-generated insights, and organizes it using hierarchical graph community detection and summarization for precise and contextually relevant information retrieval. Our key innovations include: (1) a dense medical knowledge structuring approach enabling accurate retrieval of relevant information; (2) a dynamic knowledge retrieval mechanism that enriches patient contexts with focused, multi-faceted medical insights; and (3) a reasoning-enhanced prediction framework that leverages these enriched contexts to produce both accurate and interpretable clinical predictions. Extensive experiments demonstrate that KARE outperforms leading models by up to 10.8-15.0\% on MIMIC-III and 12.6-12.7\% on MIMIC-IV for mortality and readmission predictions. In addition to its impressive prediction accuracy, our framework leverages the reasoning capabilities of LLMs, enhancing the trustworthiness of clinical predictions.

KDD ’25, August 3–7, 2025, Toronto, ON, Canada

Shuyang Yu, Runxue Bao, Parminder Bhatia, Taha Kass-Hout, Jiayu Zhou, Cao Xiao. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Marked temporal point processes (MTPPs) model sequences of events occurring at irregular time intervals, with wide-ranging applications in fields such as healthcare, finance and social networks. We propose the _state-space point process_ (S2P2) model, a novel and performant model that leverages techniques derived for modern deep state-space models (SSMs) to overcome limitations of existing MTPP models, while simultaneously imbuing strong inductive biases for continuous-time event sequences that other discrete sequence models (i.e., RNNs, transformers) do not capture. Inspired by the classical linear Hawkes processes, we propose an architecture that interleaves stochastic jump differential equations with nonlinearities to create a highly expressive intensity-based MTPP model, without the need for restrictive parametric assumptions for the intensity. Our approach enables efficient training and inference with a parallel scan, bringing linear complexity and sublinear scaling while retaining expressivity to MTPPs. Empirically, S2P2 achieves state-of-the-art predictive likelihoods across eight real-world datasets, delivering an average improvement of 33% over the best existing approaches.

LLM-as-a-Judge has emerged as a promising tool for automatically evaluating generated outputs, but its reliability is often undermined by potential biases in judgment. Existing efforts to mitigate these biases face key limitations: in-context learning-based methods fail to address rooted biases due to the evaluator’s limited capacity for self-reflection, whereas fine-tuning is not applicable to all evaluator types, especially closed-source models. To address this challenge, we introduce the **R**easoning-based **B**ias **D**etector (RBD), which is a plug-in module that identifies biased evaluations and generates structured reasoning to guide evaluator self-correction. Rather than modifying the evaluator itself, RBD operates externally and engages in an iterative process of bias detection and feedback-driven revision. To support its development, we design a complete pipeline consisting of biased dataset construction, supervision collection, distilled reasoning-based fine-tuning of RBD, and integration with LLM evaluators. We fine-tune four sizes of RBD models, ranging from 1.5B to 14B, and observe consistent performance improvements across all scales. Experimental results on 4 bias types—verbosity, position, bandwagon, and sentiment—evaluated using 8 LLM evaluators demonstrate RBD’s strong effectiveness. For example, the RBD-8B model improves evaluation accuracy by an average of 18.5% and consistency by 10.9%, and surpasses prompting-based baselines and fine-tuned judges by 12.8% and 17.2%, respectively. These results highlight RBD’s effectiveness and scalability. Additional experiments further demonstrate its strong generalization across biases and domains, as well as its efficiency.

OBJECTIVE: The objective was to develop and operate a cloud-based federated system for managing, analyzing, and sharing patient data for research purposes, while allowing each resource sharing patient data to operate their component based upon their own governance rules. The federated system is called the Biomedical Research Hub (BRH). MATERIALS AND METHODS: The BRH is a cloud-based federated system built over a core set of software services called framework services. BRH framework services include authentication and authorization, services for generating and assessing findable, accessible, interoperable, and reusable (FAIR) data, and services for importing and exporting bulk clinical data. The BRH includes data resources providing data operated by different entities and workspaces that can access and analyze data from one or more of the data resources in the BRH. RESULTS: The BRH contains multiple data commons that in aggregate provide access to over 6 PB of research data from over 400 000 research participants. DISCUSSION AND CONCLUSION: With the growing acceptance of using public cloud computing platforms for biomedical research, and the growing use of opaque persistent digital identifiers for datasets, data objects, and other entities, there is now a foundation for systems that federate data from multiple independently operated data resources that expose FAIR application programming interfaces, each using a separate data model. Applications can be built that access data from one or more of the data resources.

OBJECTIVES: This article summarizes past and current data mining activities at the United States Food and Drug Administration (FDA). TARGET AUDIENCE: We address data miners in all sectors, anyone interested in the safety of products regulated by the FDA (predominantly medical products, food, veterinary products and nutrition, and tobacco products), and those interested in FDA activities. SCOPE: Topics include routine and developmental data mining activities, short descriptions of mined FDA data, advantages and challenges of data mining at the FDA, and future directions of data mining at the FDA.

OBJECTIVE: The objective of openFDA is to facilitate access and use of big important Food and Drug Administration public datasets by developers, researchers, and the public through harmonization of data across disparate FDA datasets provided via application programming interfaces (APIs). MATERIALS AND METHODS: Using cutting-edge technologies deployed on FDA's new public cloud computing infrastructure, openFDA provides open data for easier, faster (over 300 requests per second per process), and better access to FDA datasets; open source code and documentation shared on GitHub for open community contributions of examples, apps and ideas; and infrastructure that can be adopted for other public health big data challenges. RESULTS: Since its launch on June 2, 2014, openFDA has developed four APIs for drug and device adverse events, recall information for all FDA-regulated products, and drug labeling. There have been more than 20 million API calls (more than half from outside the United States), 6000 registered users, 20,000 connected Internet Protocol addresses, and dozens of new software (mobile or web) apps developed. A case study demonstrates a use of openFDA data to understand an apparent association of a drug with an adverse event. CONCLUSION: With easier and faster access to these datasets, consumers worldwide can learn more about FDA-regulated products.

Many public health agencies monitor population health using syndromic surveillance, generally employing information from emergency department (ED) visit records. When combined with other information, objective evidence of fever may enhance the accuracy with which surveillance systems detect syndromes of interest, such as influenza-like illness. This study found that patient chief complaint of self-reported fever was more readily available in ED records than measured temperature and that the majority of patients with an elevated temperature recorded also self-reported fever. Due to its currently limited availability, we conclude that measured temperature is likely to add little value to self-reported fever in syndromic surveillance for febrile illness using ED records.

BACKGROUND: The utility of healthcare utilization data from US emergency departments (EDs) for rapid monitoring of changes in influenza-like illness (ILI) activity was highlighted during the recent influenza A (H1N1) pandemic. Monitoring has tended to rely on detection algorithms, such as the Early Aberration Reporting System (EARS), which are limited in their ability to detect subtle changes and identify disease trends. OBJECTIVE: To evaluate a complementary approach, change point analysis (CPA), for detecting changes in the incidence of ED visits due to ILI. METHODOLOGY AND PRINCIPAL FINDINGS: Data collected through the Distribute project (isdsdistribute.org), which aggregates data on ED visits for ILI from over 50 syndromic surveillance systems operated by state or local public health departments were used. The performance was compared of the cumulative sum (CUSUM) CPA method in combination with EARS and the performance of three CPA methods (CUSUM, structural change model and Bayesian) in detecting change points in daily time-series data from four contiguous US states participating in the Distribute network. Simulation data were generated to assess the impact of autocorrelation inherent in these time-series data on CPA performance. The CUSUM CPA method was robust in detecting change points with respect to autocorrelation in time-series data (coverage rates at 90% when -0.2≤ρ≤0.2 and 80% when -0.5≤ρ≤0.5). During the 2008-9 season, 21 change points were detected and ILI trends increased significantly after 12 of these change points and decreased nine times. In the 2009-10 flu season, we detected 11 change points and ILI trends increased significantly after two of these change points and decreased nine times. Using CPA combined with EARS to analyze automatically daily ED-based ILI data, a significant increase was detected of 3% in ILI on April 27, 2009, followed by multiple anomalies in the ensuing days, suggesting the onset of the H1N1 pandemic in the four contiguous states. CONCLUSIONS AND SIGNIFICANCE: As a complementary approach to EARS and other aberration detection methods, the CPA method can be used as a tool to detect subtle changes in time-series data more effectively and determine the moving direction (ie, up, down, or stable) in ILI trends between change points. The combined use of EARS and CPA might greatly improve the accuracy of outbreak detection in syndromic surveillance systems.