❌

Normal view

Received — 21 January 2026 ⏭ Favourites

SciHorizon-GENE: Benchmarking LLM for Life Sciences Inference from Gene Knowledge to Functional Understanding

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xiaohan Huang · Meng Xiao · Chuan Qin · Qingqing Long · Jinmiao Chen · Yuanchun Zhou · Hengshu Zhu
21 January 2026 at 13:00
arXiv:2601.12805v1 Announce Type: cross Abstract: Large language models (LLMs) have shown growing promise in biomedical research, particularly for knowledge-driven interpretation tasks. However, their ability to reliably reason from gene-level knowledge to functional understanding, However, their ability to reliably reason from gene-level knowledge to functional understanding, a core requirement for knowledge-enhanced cell atlas interpretation, remains largely underexplored. To address this gap, we introduce SciHorizon-GENE, a large-scale gene-centric benchmark constructed from authoritative biological databases. The benchmark integrates curated knowledge for over 190K human genes and comprises more than 540K questions covering diverse gene-to-function reasoning scenarios relevant to cell type annotation, functional interpretation, and mechanism-oriented analysis. Motivated by behavioral patterns observed in preliminary examinations, SciHorizon-GENE evaluates LLMs along four biologically critical perspectives: research attention sensitivity, hallucination tendency, answer completeness, and literature influence, explicitly targeting failure modes that limit the safe adoption of LLMs in biological interpretation pipelines. We systematically evaluate a wide range of state-of-the-art general-purpose and biomedical LLMs, revealing substantial heterogeneity in gene-level reasoning capabilities and persistent challenges in generating faithful, complete, and literature-grounded functional interpretations. Our benchmark establishes a systematic foundation for analyzing LLM behavior at the gene scale and offers insights for model selection and development, with direct relevance to knowledge-enhanced biological interpretation.

Zero-shot adaptable task planning for autonomous construction robots: a comparative study of lightweight single and multi-AI agent systems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Hossein Naderi · Alireza Shojaei · Lifu Huang · Philip Agee · Kereshmeh Afsari · Abiola Akanmu
21 January 2026 at 13:00
arXiv:2601.14091v1 Announce Type: cross Abstract: Robots are expected to play a major role in the future construction industry but face challenges due to high costs and difficulty adapting to dynamic tasks. This study explores the potential of foundation models to enhance the adaptability and generalizability of task planning in construction robots. Four models are proposed and implemented using lightweight, open-source large language models (LLMs) and vision language models (VLMs). These models include one single agent and three multi-agent teams that collaborate to create robot action plans. The models are evaluated across three construction roles: Painter, Safety Inspector, and Floor Tiling. Results show that the four-agent team outperforms the state-of-the-art GPT-4o in most metrics while being ten times more cost-effective. Additionally, teams with three and four agents demonstrate the improved generalizability. By discussing how agent behaviors influence outputs, this study enhances the understanding of AI teams and supports future research in diverse unstructured environments beyond construction.

OpenNovelty: An LLM-powered Agentic System for Verifiable Scholarly Novelty Assessment

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ming Zhang · Kexin Tan · Yueyuan Huang · Yujiong Shen · Chunchun Ma · Li Ju · Xinran Zhang · Yuhui Wang · Wenqing Jing · Jingyi Deng · Huayu Sha · Binze Hu · Jingqi Tong · Changhao Jiang · Yage Geng · Yuankai Ying · Yue Zhang · Zhangyue Yin · Zhiheng Xi · Shihan Dou · Tao Gui · Qi Zhang · Xuanjing Huang
21 January 2026 at 13:00
arXiv:2601.01576v2 Announce Type: replace-cross Abstract: Evaluating novelty is critical yet challenging in peer review, as reviewers must assess submissions against a vast, rapidly evolving literature. This report presents OpenNovelty, an LLM-powered agentic system for transparent, evidence-based novelty analysis. The system operates through four phases: (1) extracting the core task and contribution claims to generate retrieval queries; (2) retrieving relevant prior work based on extracted queries via semantic search engine; (3) constructing a hierarchical taxonomy of core-task-related work and performing contribution-level full-text comparisons against each contribution; and (4) synthesizing all analyses into a structured novelty report with explicit citations and evidence snippets. Unlike naive LLM-based approaches, \textsc{OpenNovelty} grounds all assessments in retrieved real papers, ensuring verifiable judgments. We deploy our system on 500+ ICLR 2026 submissions with all reports publicly available on our website, and preliminary analysis suggests it can identify relevant prior work, including closely related papers that authors may overlook. OpenNovelty aims to empower the research community with a scalable tool that promotes fair, consistent, and evidence-backed peer review.
Received — 16 January 2026 ⏭ Favourites

Mitigating Gender Bias via Fostering Exploratory Thinking in LLMs

✇cs.AI, q-bio.NC updates on arXiv.org
By: Kangda Wei · Hasnat Md Abdullah · Ruihong Huang
15 January 2026 at 13:00
arXiv:2505.17217v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) often exhibit gender bias, resulting in unequal treatment of male and female subjects across different contexts. To address this issue, we propose a novel data generation framework that fosters exploratory thinking in LLMs. Our approach prompts models to generate story pairs featuring male and female protagonists in structurally identical, morally ambiguous scenarios, then elicits and compares their moral judgments. When inconsistencies arise, the model is guided to produce balanced, gender-neutral judgments. These story-judgment pairs are used to fine-tune or optimize the models via Direct Preference Optimization (DPO). Experimental results show that our method significantly reduces gender bias while preserving or even enhancing general model capabilities. We will release the code and generated data. We release the code and generated data at: https://github.com/WeiKangda/LLMs-Exploratory-Bias-Mitigation/tree/main.

Complement-secreting CAFs are associated with better prognosis in pancreatic cancer: single-cell multiomics

✇(Multiomics OR Omics) AND (Pancreatic)
By: Kai Chen · Yongsu Ma · Liling Huang · Pengfei Wu · Heng-Chung Kung · Bohan Yang · Jixin Zhang · Robert A Anders · Jacquelyn W Zimmerman · Qingfeng Zhu · Xiaodong Tian · Jin He · Yinmo Yang
14 January 2026 at 19:00

Gut. 2026 Jan 13:gutjnl-2025-335683. doi: 10.1136/gutjnl-2025-335683. Online ahead of print.

ABSTRACT

BACKGROUND: Accumulating evidence has demonstrated that distinct tumour-promoting and tumour-restraining cancer-associated fibroblast (CAF) subtypes coexist in pancreatic ductal adenocarcinoma.

OBJECTIVE: To develop targeted CAF therapeutic strategies by reprogramming tumour-promoting CAF subtypes.

DESIGN: We leveraged multiomics technologies to systematically identify and characterise CAF subtypes transcriptionally, epigenetically and spatially and correlate them with clinicopathological features.

RESULTS: We found that complement-secreting CAFs (csCAFs), initially identified by our group and inflammatory CAFs (iCAFs) share significant overlap in their transcriptional profiles and chromatin accessibility. iCAFs specifically express transcription factors from the heme and oxidative homeostasis pathway and the activator protein 1 family, which are both involved in cellular response to oxidative stress. Notably, the composition of csCAFs among all CAFs declined during pancreatic carcinogenesis, while trajectory analysis showed that csCAFs could potentially differentiate into iCAFs. Spatially resolved analysis indicated that tumour regions with a higher csCAF composition were associated with lower levels of TGF-β ligands, fewer M2 tumour-associated macrophages and increased levels of lipid mediators. Additionally, we identified a spatially defined CXCL12-CXCR4 ligand-receptor interaction between csCAFs and T cells, but in distinct patterns between different metastatic organs. Patients with a higher composition of csCAFs have significantly longer overall survival and recurrence-free survival through multiplex immunohistochemistry and bulk RNA-seq deconvolution.

CONCLUSION: Our study demonstrates that csCAFs may represent an early-stage iCAF subtype and suggests a promising strategy for reprogramming iCAFs into csCAFs.

PMID:41534892 | DOI:10.1136/gutjnl-2025-335683

Received — 14 January 2026 ⏭ Favourites
Received — 13 January 2026 ⏭ Favourites

Interoperability in AI Safety Governance: Ethics, Regulations, and Standards

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yik Chan Chin · David A. Raho · Hag-Min Kim · Chunli Bi · James Ong · Jingbo Huang · Serge Stinckwich
13 January 2026 at 13:00
arXiv:2601.06153v1 Announce Type: cross Abstract: This policy report draws on country studies from China, South Korea, Singapore, and the United Kingdom to identify effective tools and key barriers to interoperability in AI safety governance. It offers practical recommendations to support a globally informed yet locally grounded governance ecosystem. Interoperability is a central goal of AI governance, vital for reducing risks, fostering innovation, enhancing competitiveness, promoting standardization, and building public trust. However, structural gaps such as fragmented regulations and lack of global coordination, and conceptual gaps, including limited Global South engagement, continue to hinder progress. Focusing on three high-stakes domains - autonomous vehicles, education, and cross-border data flows - the report compares ethical, legal, and technical frameworks across the four countries. It identifies areas of convergence, divergence, and potential alignment, offering policy recommendations that support the development of interoperability mechanisms aligned with the Global Digital Compact and relevant UN resolutions. The analysis covers seven components: objectives, regulators, ethics, binding measures, targeted frameworks, technical standards, and key risks.

FairMedQA: Benchmarking Bias in Large Language Models for Medical Question Answering

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ying Xiao · Jie Huang · Ruijuan He · Jing Xiao · Mohammad Reza Mousavi · Yepang Liu · Kezhi Li · Zhenpeng Chen · Jie M. Zhang
13 January 2026 at 13:00
arXiv:2505.19562v2 Announce Type: replace Abstract: Large language models (LLMs) are approaching expert-level performance in medical question answering (QA), demonstrating strong potential to improve public healthcare. However, underlying biases related to sensitive attributes such as sex and race pose life-critical risks. The extent to which such sensitive attributes affect diagnosis remains an open question and requires comprehensive empirical investigation. Additionally, even the latest Counterfactual Patient Variations (CPV) benchmark can hardly distinguish the bias levels of different LLMs. To further explore these dynamics, we propose a new benchmark, FairMedQA, and benchmark 12 representative LLMs. FairMedQA contains 4,806 counterfactual question pairs constructed from 801 clinical vignettes. Our results reveal substantial accuracy disparity ranging from 3 to 19 percentage points across sensitive demographic groups. Notably, FairMedQA exposes biases that are at least 12 percentage points larger than those identified by the latest CPV benchmark, presenting superior benchmarking sensitivity. Our results underscore an urgent need for targeted debiasing techniques and more rigorous, identity-aware validation protocols before LLMs can be safely integrated into practical clinical decision-support systems.
Received — 12 January 2026 ⏭ Favourites

Streamlining evidence based clinical recommendations with large language models

✇cs.AI, q-bio.NC updates on arXiv.org
By: Dubai Li · Nan Jiang · Kangping Huang · Ruiqi Tu · Shuyu Ouyang · Huayu Yu · Lin Qiao · Chen Yu · Tianshu Zhou · Danyang Tong · Qian Wang · Mengtao Li · Xiaofeng Zeng · Yu Tian · Xinping Tian · Jingsong Li
12 January 2026 at 13:00
arXiv:2505.10282v2 Announce Type: replace-cross Abstract: Clinical evidence underpins informed healthcare decisions, yet integrating it into real-time practice remains challenging due to intensive workloads, complex procedures, and time constraints. This study presents Quicker, an LLM-powered system that automates evidence synthesis and generates clinical recommendations following standard guideline development workflows. Quicker delivers an end-to-end pipeline from clinical questions to recommendations and supports customized decision-making through integrated tools and interactive interfaces. To evaluate how closely Quicker can reproduce guideline development processes, we constructed Q2CRBench-3, a benchmark derived from guideline development records for three diseases. Experiments show that Quicker produces precise question decomposition, expert-aligned retrieval, and near-comprehensive screening. Quicker assistance improved the accuracy of extracted study data, and its recommendations were more comprehensive and coherent than clinician-written ones. In system-level testing, Quicker working with one participant reduced recommendation development to 20-40 min. Overall, the findings demonstrate Quicker's potential to enhance the speed and reliability of evidence-based clinical decision-making.
Received — 7 January 2026 ⏭ Favourites

The role of PCMT1 in prognosis tumor immune microenvironment and therapeutic responses across cancers

✇Omics In Lung
By: Bo Wang · Sijia Huang · Ruizhen Ren · Ruiqian Yao · Erwen Kou · Haixia Zhao · Hao Zhu · Mengyu Zhang · Liangzhe Wang · Yuanjie Zhu
6 January 2026 at 19:00

Discov Oncol. 2026 Jan 5. doi: 10.1007/s12672-025-04366-2. Online ahead of print.

ABSTRACT

BACKGROUND: Emerging evidence highlights the overexpression of Protein-L-isoaspartate (D-aspartate) O-methyltransferase (PCMT1) in multiple malignancies. However, its pan-cancer prognostic significance, tumor immune microenvironment (TIME) interactions, and therapeutic implications remain underexplored.

METHODS: Multi-omics data were integrated from UCSC Xena, GTEx, UALCAN, and published cohorts. PCMT1 expression patterns were systematically analyzed across 33 cancer types. Associations between PCMT1 and clinical outcomes, immune infiltration, immune checkpoint genes (ICGs), tumor mutation burden (TMB), microsatellite instability (MSI), and drug sensitivity were evaluated using bioinformatics pipelines.

RESULTS: Our pan-cancer analysis revealed differential expression patterns of PCMT1 across various malignancies, with significant upregulation in 20 cancer types and downregulation in 3 cancer types. Notably, PCMT1 overexpression was predominantly observed in epithelial-origin tumors, such as ACC (adrenocortical carcinoma), BRCA (breast invasive carcinoma), COAD (colon adenocarcinoma), and LUAD (lung adenocarcinoma). Survival analysis demonstrated that elevated PCMT1 expression was significantly correlated with unfavorable prognosis in multiple epithelial tumors, particularly in BRCA, esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), liver hepatocellular carcinoma (LIHC), and mesothelioma (MESO). Furthermore, comprehensive analysis identified significant associations between PCMT1 expression and various tumor microenvironment features, including immune scores, six distinct immune cell types, four immunosuppressive cell populations, cancer-associated fibroblasts (CAFs)-related markers, and immunosuppressive factors. PCMT1 expression also showed significant correlations with tumor mutation burden (TMB), microsatellite instability (MSI), DNA stemness score (DNAss), and RNA stemness score (RNAss). Particularly noteworthy was the strong positive correlation between PCMT1 expression and CAFs infiltration, along with their associated factors. These findings were further validated in independent immunotherapy cohorts, where PCMT1 consistently demonstrated immunosuppressive characteristics.

CONCLUSION: Multi-omics analysis suggests that PCMT1 may serve as a potential prognostic biomarker and a novel immunotherapy target for pan-cancer.

PMID:41491065 | DOI:10.1007/s12672-025-04366-2

Digital Twin AI: Opportunities and Challenges from Large Language Models to World Models

✇cs.AI, q-bio.NC updates on arXiv.org
By: Rong Zhou · Dongping Chen · Zihan Jia · Yao Su · Yixin Liu · Yiwen Lu · Dongwei Shi · Yue Huang · Tianyang Xu · Yi Pan · Xinliang Li · Yohannes Abate · Qingyu Chen · Zhengzhong Tu · Yu Yang · Yu Zhang · Qingsong Wen · Gengchen Mai · Sunyang Fu · Jiachen Li · Xuyu Wang · Ziran Wang · Jing Huang · Tianming Liu · Yong Chen · Lichao Sun · Lifang He
6 January 2026 at 13:00
arXiv:2601.01321v1 Announce Type: new Abstract: Digital twins, as precise digital representations of physical systems, have evolved from passive simulation tools into intelligent and autonomous entities through the integration of artificial intelligence technologies. This paper presents a unified four-stage framework that systematically characterizes AI integration across the digital twin lifecycle, spanning modeling, mirroring, intervention, and autonomous management. By synthesizing existing technologies and practices, we distill a unified four-stage framework that systematically characterizes how AI methodologies are embedded across the digital twin lifecycle: (1) modeling the physical twin through physics-based and physics-informed AI approaches, (2) mirroring the physical system into a digital twin with real-time synchronization, (3) intervening in the physical twin through predictive modeling, anomaly detection, and optimization strategies, and (4) achieving autonomous management through large language models, foundation models, and intelligent agents. We analyze the synergy between physics-based modeling and data-driven learning, highlighting the shift from traditional numerical solvers to physics-informed and foundation models for physical systems. Furthermore, we examine how generative AI technologies, including large language models and generative world models, transform digital twins into proactive and self-improving cognitive systems capable of reasoning, communication, and creative scenario generation. Through a cross-domain review spanning eleven application domains, including healthcare, aerospace, smart manufacturing, robotics, and smart cities, we identify common challenges related to scalability, explainability, and trustworthiness, and outline directions for responsible AI-driven digital twin systems.

OpenNovelty: An LLM-powered Agentic System for Verifiable Scholarly Novelty Assessment

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ming Zhang · Kexin Tan · Yueyuan Huang · Yujiong Shen · Chunchun Ma · Li Ju · Xinran Zhang · Yuhui Wang · Wenqing Jing · Jingyi Deng · Huayu Sha · Binze Hu · Jingqi Tong · Changhao Jiang · Yage Geng · Yuankai Ying · Yue Zhang · Zhangyue Yin · Zhiheng Xi · Shihan Dou · Tao Gui · Qi Zhang · Xuanjing Huang
6 January 2026 at 13:00
arXiv:2601.01576v1 Announce Type: cross Abstract: Evaluating novelty is critical yet challenging in peer review, as reviewers must assess submissions against a vast, rapidly evolving literature. This report presents OpenNovelty, an LLM-powered agentic system for transparent, evidence-based novelty analysis. The system operates through four phases: (1) extracting the core task and contribution claims to generate retrieval queries; (2) retrieving relevant prior work based on extracted queries via semantic search engine; (3) constructing a hierarchical taxonomy of core-task-related work and performing contribution-level full-text comparisons against each contribution; and (4) synthesizing all analyses into a structured novelty report with explicit citations and evidence snippets. Unlike naive LLM-based approaches, \textsc{OpenNovelty} grounds all assessments in retrieved real papers, ensuring verifiable judgments. We deploy our system on 500+ ICLR 2026 submissions with all reports publicly available on our website, and preliminary analysis suggests it can identify relevant prior work, including closely related papers that authors may overlook. OpenNovelty aims to empower the research community with a scalable tool that promotes fair, consistent, and evidence-backed peer review.
Received — 30 December 2025 ⏭ Favourites

PathFound: An Agentic Multimodal Model Activating Evidence-seeking Pathological Diagnosis

✇cs.AI, q-bio.NC updates on arXiv.org
By: Shengyi Hua · Jianfeng Wu · Tianle Shen · Kangzhe Hu · Zhongzhen Huang · Shujuan Ni · Zhihong Zhang · Yuan Li · Zhe Wang · Xiaofan Zhang
30 December 2025 at 13:00
arXiv:2512.23545v1 Announce Type: cross Abstract: Recent pathological foundation models have substantially advanced visual representation learning and multimodal interaction. However, most models still rely on a static inference paradigm in which whole-slide images are processed once to produce predictions, without reassessment or targeted evidence acquisition under ambiguous diagnoses. This contrasts with clinical diagnostic workflows that refine hypotheses through repeated slide observations and further examination requests. We propose PathFound, an agentic multimodal model designed to support evidence-seeking inference in pathological diagnosis. PathFound integrates the power of pathological visual foundation models, vision-language models, and reasoning models trained with reinforcement learning to perform proactive information acquisition and diagnosis refinement by progressing through the initial diagnosis, evidence-seeking, and final decision stages. Across several large multimodal models, adopting this strategy consistently improves diagnostic accuracy, indicating the effectiveness of evidence-seeking workflows in computational pathology. Among these models, PathFound achieves state-of-the-art diagnostic performance across diverse clinical scenarios and demonstrates strong potential to discover subtle details, such as nuclear features and local invasions.
Received — 28 December 2025 ⏭ Favourites

Context matching is not reasoning when performing generalized clinical evaluation of generative language models

✇npj Digital Medicine
By: Andrew Wen · Qiuhao Lu · Yu-Neng Chuang · Guanchu Wang · Jiayi Yuan · Jiamu Zhang · Liwei Wang · Sunyang Fu · Kurt D. Miller · Heling Jia · Steven D. Bedrick · William R. Hersh · Kirk E. Roberts · Xia Hu · Hongfang Liu
27 December 2025 at 08:00

npj Digital Medicine, Published online: 27 December 2025; doi:10.1038/s41746-025-02253-2

Context matching is not reasoning when performing generalized clinical evaluation of generative language models

A novel evaluation benchmark for medical LLMs illuminating safety and effectiveness in clinical domains

✇npj Digital Medicine
By: Shirui Wang · Zhihui Tang · Huaxia Yang · Qiuhong Gong · Tiantian Gu · Hongyang Ma · Yongxin Wang · Wubin Sun · Zeliang Lian · Kehang Mao · Yinan Jiang · Zhicheng Huang · Lingyun Ma · Wenjie Shen · Yajie Ji · Yunhui Tan · Chunbo Wang · Yunlu Gao · Qianling Ye · Rui Lin · Mingyu Chen · Lijuan Niu · Zhihao Wang · Peng Yu · Mengran Lang · Yue Liu · Huimin Zhang · Haitao Shen · Long Chen · Qiguang Zhao · Si-Xuan Liu · Lina Zhou · Hua Gao · Dongqiang Ye · Lingmin Meng · Youtao Yu · Naixin Liang · Jianxiong Wu
26 December 2025 at 08:00

npj Digital Medicine, Published online: 26 December 2025; doi:10.1038/s41746-025-02277-8

A novel evaluation benchmark for medical LLMs illuminating safety and effectiveness in clinical domains
Received — 18 December 2025 ⏭ Favourites

aiXiv: A Next-Generation Open Access Ecosystem for Scientific Discovery Generated by AI Scientists

✇cs.AI, q-bio.NC updates on arXiv.org
By: Pengsong Zhang · Xiang Hu · Guowei Huang · Yang Qi · Heng Zhang · Xiuxu Li · Jiaxing Song · Jiabin Luo · Yijiang Li · Shuo Yin · Chengxiao Dai · Eric Hanchen Jiang · Xiaoyan Zhou · Zhenfei Yin · Boqin Yuan · Jing Dong · Guinan Su · Guanren Qiao · Haiming Tang · Anghong Du · Lili Pan · Zhenzhong Lan · Xinyu Liu
18 December 2025 at 13:00
arXiv:2508.15126v2 Announce Type: replace Abstract: Recent advances in large language models (LLMs) have enabled AI agents to autonomously generate scientific proposals, conduct experiments, author papers, and perform peer reviews. Yet this flood of AI-generated research content collides with a fragmented and largely closed publication ecosystem. Traditional journals and conferences rely on human peer review, making them difficult to scale and often reluctant to accept AI-generated research content; existing preprint servers (e.g. arXiv) lack rigorous quality-control mechanisms. Consequently, a significant amount of high-quality AI-generated research lacks appropriate venues for dissemination, hindering its potential to advance scientific progress. To address these challenges, we introduce aiXiv, a next-generation open-access platform for human and AI scientists. Its multi-agent architecture allows research proposals and papers to be submitted, reviewed, and iteratively refined by both human and AI scientists. It also provides API and MCP interfaces that enable seamless integration of heterogeneous human and AI scientists, creating a scalable and extensible ecosystem for autonomous scientific discovery. Through extensive experiments, we demonstrate that aiXiv is a reliable and robust platform that significantly enhances the quality of AI-generated research proposals and papers after iterative revising and reviewing on aiXiv. Our work lays the groundwork for a next-generation open-access ecosystem for AI scientists, accelerating the publication and dissemination of high-quality AI-generated research content. Code: https://github.com/aixiv-org aiXiv: https://aixiv.science

Why Text Prevails: Vision May Undermine Multimodal Medical Decision Making

✇cs.AI, q-bio.NC updates on arXiv.org
By: Siyuan Dai · Lunxiao Li · Kun Zhao · Eardi Lila · Paul K. Crane · Heng Huang · Dongkuan Xu · Haoteng Tang · Liang Zhan
17 December 2025 at 13:00
arXiv:2512.13747v1 Announce Type: cross Abstract: With the rapid progress of large language models (LLMs), advanced multimodal large language models (MLLMs) have demonstrated impressive zero-shot capabilities on vision-language tasks. In the biomedical domain, however, even state-of-the-art MLLMs struggle with basic Medical Decision Making (MDM) tasks. We investigate this limitation using two challenging datasets: (1) three-stage Alzheimer's disease (AD) classification (normal, mild cognitive impairment, dementia), where category differences are visually subtle, and (2) MIMIC-CXR chest radiograph classification with 14 non-mutually exclusive conditions. Our empirical study shows that text-only reasoning consistently outperforms vision-only or vision-text settings, with multimodal inputs often performing worse than text alone. To mitigate this, we explore three strategies: (1) in-context learning with reason-annotated exemplars, (2) vision captioning followed by text-only inference, and (3) few-shot fine-tuning of the vision tower with classification supervision. These findings reveal that current MLLMs lack grounded visual understanding and point to promising directions for improving multimodal decision making in healthcare.

Automated Multitier Tagging of Chinese Online Health Education Resources Using a Large Language Model: Development and Validation Study

✇Journal of Medical Internet Research
By: Jialin Meng · Ruiming Dai · Xiaolan Huang · Yi Gu · Shixing Yan · Xiaoke Wang · Jingrong Gao · Tian-Tian Zhang
18 December 2025 at 06:00
Background: Precision health promotion, which aims to tailor health messages to individual needs, is hampered by the lack of structured metadata in vast digital health resource libraries. This bottleneck prevents scalable, personalized content delivery and exacerbates information overload for the public. Objective: This study aimed to develop, deploy, and validate an automated tagging system using a large language model (LLM) to create the foundational metadata infrastructure required for tailored health communication at scale. Methods: We developed a comprehensive, 3-tier health promotion taxonomy (10 primary, 34 secondary, and 90,562 tertiary tags) using a hybrid Delphi and corpus-mining methodology. We then constructed a hybrid inference pipeline by fine-tuning a Baichuan2-7B LLM with low-rank adaptation for initial tag generation. This was then refined by a domain-specific named entity recognition model and standardized against a vector database. The system’s performance was evaluated against manual annotations from nonexpert staff on a test set of 1000 resources. We used a “no gold standard” framework, comparing the artificial intelligence–human (A-H) interrater reliability (IRR) with a supplemental human-human (H-H) IRR baseline and expert adjudication for cases where artificial intelligence provided additional tags (“AI Additive”). Results: The A-H agreement was moderate (Cohen κ=0.54, 95% CI 0.53-0.56; Jaccard similarity coefficient=0.48, 95% CI 0.46-0.50). Critically, this was higher than the baseline nonexpert H-H agreement (Cohen κ=0.32, 95% CI 0.29-0.35; Jaccard similarity coefficient=0.35, 95% CI 0.27-0.43). A granular analysis of disagreements revealed that in 15.9% (159/1000) of the cases, the “AI Additive” tags were not identified by human annotators. Expert adjudication of these cases confirmed that the “AI Additive” tags were correct and relevant with a precision of 90% (45/50; 95% CI 78.2%-96.7%). Conclusions: A fine-tuned LLM, integrated into a hybrid pipeline, can function as a powerful augmentation tool for health content annotation. The system’s consistency (A-H κ=0.54) was found to be superior to the baseline human workflow (H-H κ=0.32). By moving beyond simple automation to reliably identify relevant health topics missed by manual annotators with high, expert-validated accuracy, this study provides a robust technical and methodological blueprint for implementing artificial intelligence to enhance precision health communication in public health settings.
Received — 14 December 2025 ⏭ Favourites

Exploring the role of lipid metabolism genes in gastric cancer prognosis and tumor immune microenvironment

✇(Multiomics OR Omics) AND (Lung OR gastric OR Hepatocellular)
By: Yanan Xiao · Qin Yu · Zewen Zhu · Shan Zhou · Zhe Wu · Xin Luo · Xingzhi Huang · Zhiqiang Zhan
11 December 2025 at 19:00

J Int Med Res. 2025 Dec;53(12):3000605251403252. doi: 10.1177/03000605251403252. Epub 2025 Dec 11.

ABSTRACT

BackgroundGastric cancer remains a major global health challenge due to its high mortality rate and complex pathophysiological mechanisms. Emerging evidence highlights that dysregulated lipid metabolism contributes to gastric cancer progression and prognosis, but the associations between lipid metabolism-associated genes, gastric cancer patient survival, and tumor immune microenvironment remodeling are not fully elucidated.MethodsWe analyzed publicly available omics and clinical data, including RNA sequencing data from 371 gastric cancer samples in The Cancer Genome Atlas database and 433 gastric cancer samples in the Gene Expression Omnibus database. We first curated the top 100 lipid metabolism-associated genes based on relevance scores. Then, univariate Cox regression was used to identify genes significantly associated with overall survival. Consensus clustering was applied to these survival-related genes to define gastric cancer molecular subtypes. Copy number variation analysis was performed to assess genomic alterations of these genes in tumor samples. A prognostic risk model was constructed using least absolute shrinkage and selection operator regression and validated via multivariate Cox regression. Immune infiltration analysis using CIBERSORT and ESTIMATE algorithms was conducted to explore associations between lipid metabolism-associated genes and tumor immune microenvironment characteristics.ResultsA total of 3911 differentially expressed genes were identified between gastric cancer and adjacent normal tissues. Among the top 100 lipid metabolism-associated genes, 43 were significantly linked to patient survival, most of which were considered as poor prognostic factors. Copy number variation analysis revealed frequent copy number gains of these genes in tumor samples. Consensus clustering stratified patients into two molecular subtypes (LMAGcluster A and LMAGcluster B), with LMAGcluster A showing significantly worse survival outcomes (median survival: 2.6 years vs. 8.3 years in LMAGcluster B, p < 0.001). LMAGcluster A was also characterized by elevated infiltration of pro-tumor immune cells, such as regulatory T cells and follicular helper T cells. The prognostic model based on 14 key lipid metabolism-associated genes exhibited robust predictive performance, with area under the receiver operating characteristic curve values of 0.702-0.761 in The Cancer Genome Atlas cohort and 0.621-0.638 in the Gene Expression Omnibus cohort for 1-, 3-, and 5-year survival.ConclusionLipid metabolism-associated genes are closely associated with gastric cancer prognosis and tumor immune microenvironment remodeling. The identified gene-based molecular subtypes and prognostic model provide novel insights into gastric cancer progression, and the 14 key genes may serve as potential biomarkers and therapeutic targets.

PMID:41381057 | DOI:10.1177/03000605251403252

Received — 9 December 2025 ⏭ Favourites

The AI Productivity Index (APEX)

✇cs.AI, q-bio.NC updates on arXiv.org
By: Bertie Vidgen · Abby Fennelly · Evan Pinnix · Julien Benchek · Daniyal Khan · Zach Richards · Austin Bridges · Calix Huang · Ben Hunsberger · Isaac Robinson · Akul Datta · Chirag Mahapatra · Dominic Barton · Cass R. Sunstein · Eric Topol · Brendan Foody · Osvald Nitski
8 December 2025 at 13:00
arXiv:2509.25721v4 Announce Type: replace-cross Abstract: We present an extended version of the AI Productivity Index (APEX-v1-extended), a benchmark for assessing whether frontier models are capable of performing economically valuable tasks in four jobs: investment banking associate, management consultant, big law associate, and primary care physician (MD). This technical report details the extensions to APEX-v1, including an increase in the held-out evaluation set from n = 50 to n = 100 cases per job (n = 400 total) and updates to the grading methodology. We present a new leaderboard, where GPT5 (Thinking = High) remains the top performing model with a score of 67.0%. APEX-v1-extended shows that frontier models still have substantial limitations when performing typical professional tasks. To support further research, we are open sourcing n = 25 non-benchmark example cases per role (n = 100 total) along with our evaluation harness.
❌