❌

Normal view

Received — 26 October 2025 ⏭ Favourites

Decoding the tumor immune microenvironment in lung squamous cell carcinoma: characteristics, regulatory mechanisms, and future directions in immunotherapy

✇Omics In Lung
By: Yunfeng Tong · Yuxia Wang · Yunfei Chen · Yun Fan · Hui Li
24 October 2025 at 18:00

Transl Lung Cancer Res. 2025 Sep 30;14(9):4112-4130. doi: 10.21037/tlcr-2025-350. Epub 2025 Sep 18.

ABSTRACT

Lung squamous cell carcinoma (LUSC), a predominant type of lung cancer, is marked by an unfavorable prognosis and limited therapeutic options. Unlike lung adenocarcinoma (LUAD), LUSC exhibits few driver mutations, resulting in minimal benefits from targeted therapies for these patients. Despite the transformative effects of immunotherapy on patient outcomes, only a subset of patients achieving durable responses. This heterogeneity in treatment outcomes is increasingly attributed to the complex feature of the tumor immune microenvironment (TIME) in LUSC. The TIME of LUSC is a highly dynamic ecosystem composed of diverse immune cell populations and stromal components that collectively foster an immune-evasive niche. Recent breakthroughs in multi-omics technologies, particularly single-cell RNA sequencing (scRNA-seq) and spatial omics, have provided unprecedented resolution in dissecting the cellular and molecular architecture of the TIME in LUSC. These technologies have enabled the identification of distinct immune cells and their spatial interactions with the tumor, shedding light on the mechanisms underlying immune evasion and resistance to immunotherapy. Building on these advancements, this review establishes a new classification of the TIME which may guide patient stratification and personalized immunotherapy. And we comprehensively offer a detailed examination of the principal characteristics and regulatory mechanisms of the TIME, highlighting potential immunotherapeutic strategies tailored to this distinct immunological context.

PMID:41133013 | PMC:PMC12541881 | DOI:10.21037/tlcr-2025-350

Biomarkers for non-small cell lung cancer risk using multi-omics approaches: a nested case-control study

✇Omics In Lung
By: Youngmin Han · Keum Ji Jung · Seong Gyu Choi · Yeun Soo Yang · Kwangbae Lee · Sun Ha Jee
24 October 2025 at 18:00

Transl Lung Cancer Res. 2025 Sep 30;14(9):3645-3658. doi: 10.21037/tlcr-2025-603. Epub 2025 Sep 25.

ABSTRACT

BACKGROUND: Lung cancer poses a major public health challenge, accounting for the highest cancer-related mortality worldwide. This study aimed to identify non-invasive biomarkers for the early detection of non-small cell lung cancer (NSCLC) risk.

METHODS: We randomly selected 150 incident NSCLC cases during follow-up from the Korean Cancer Prevention Study-II. Controls (n=150) were matched to cases by age, gender, and the time of blood collection. Non-targeted metabolite screening by ultra-high-performance liquid chromatography (UHPLC)/mass spectrometry (MS) was conducted on the pre-diagnostic biological samples. The 11 reported lung cancer-associated single-nucleotide polymorphisms (SNPs) in Koreans were extracted from DNA genotyping data of the study population. Metabolite markers related to NSCLC risk were identified through clustering using hierarchical density-based spatial clustering of applications with noise. The associations between smoking, dietary factors, and NSCLC were also examined.

RESULTS: Six discriminative serum metabolites were identified as having an association with NSCLC incidence. Notably, the relationship between specific metabolite levels and NSCLC risk differed by rs7086803 genotype. Smoking status and occupational exposures appear to influence specific metabolite profiles, while dietary vegetable intake may modulate the risk of NSCLC among smokers.

CONCLUSIONS: The meaningful biomarkers revealed in the current research could be used to enhance the predictive ability for NSCLC risk. Furthermore, we suggest that the protective role of dietary vegetables against NSCLC may be attenuated or absent in smokers.

PMID:41133005 | PMC:PMC12541849 | DOI:10.21037/tlcr-2025-603

Best Practices for Data Modernization Across the United States Public Health System: Scoping Review

✇Journal of Medical Internet Research
By: Zebunnesa Zeba · Stella T Lartey · Polina Durneva · Shongkour Roy · Niharika Jha · Michael Arthur Ofori · Nidhi Mittal · Stella Dockery · Nichole Saulsberry Scarboro · Michelle Taylor · Ashish Joshi
25 October 2025 at 04:00
Background: The adoption of new technologies and data modernization approaches in public health aims to enhance the use of health data to inform decision-making and improve population health. However, public health departments struggle with legacy systems, siloed data, and privacy concerns, hampering new technology adoption and data sharing with stakeholders. This paper maps how to address these shortcomings by identifying data modernization challenges, initiatives, and progress. Objective: To characterize the evidence for data modernization associated gaps and best practices in public health. Methods: This scoping review was conducted using the five-stage framework developed by Arksey and O’Malley and was reported according to the PRISMA-ScR guidelines. A structured search was performed in databases PubMed, Scopus, CINAHL, PsycINFO, and was complemented by a further search in the Google Scholar search engine, covering publications from January 1, 2019, to April 30, 2024. Eligible studies were peer-reviewed, published in English, and focused on data modernization initiatives within U.S. public health and reported on best practices, challenges, and outcomes. Search terms combined concepts such as “Data Modernization,” “Interoperability,” and “Public Health” using Boolean operators. Two reviewers independently screened titles, abstracts, and full texts using Rayyan QCRI, with conflicts resolved through consultation with a third reviewer. Data was extracted into Microsoft Excel and thematically analyzed. Results: This review analyzed 22 studies focused on public health data modernization. Across the literature, common components included transitioning to cloud-based systems, consolidating fragmented data into unified platforms, applying governance frameworks, and implementing analytics tools to support decision-making. Primary data sources were electronic health records, insurance claims, and disease surveillance registries. Key challenges identified across studies involved data quality issues, lack of interoperability, and limited resources, particularly in underfunded settings. Notable benefits included more timely and accessible data, improved integration across systems, and enhanced analytical capabilities, which collectively support more responsive and effective public health interventions when guided by clear standards and policy alignment. Conclusions: Progress hinges on balancing local adaptability with national coordination, improving data governance practices, and enhancing collaboration across institutions. These steps are vital to ensure public health systems can deliver timely, accurate, and actionable information to support effective public health efforts.
  • ✇STAT
  • Opinion: How much should healthy medical research volunteers get paid? Torie Bosch
    Below is a lightly edited, AI-generated transcript of the “First Opinion Podcast” interview with Jake Eberts and Jill Fisher. Be sure to sign up for the weekly “First Opinion Podcast” on Apple Podcasts, Spotify, or wherever you get your podcasts. Get alerts about each new episode by signing up for the “First Opinion Podcast” newsletter. And don’t forget to sign up for the First Opinion newsletter, delivered every Sunday. Torie Bosch: Jake Eberts did not die of dysentery. But he did catc
     
  • 🏷️
  • ↗️

Opinion: How much should healthy medical research volunteers get paid?

✇STAT
By: Torie Bosch
25 October 2025 at 19:00

Below is a lightly edited, AI-generated transcript of the “First Opinion Podcast” interview with Jake Eberts and Jill Fisher. Be sure to sign up for the weekly “First Opinion Podcast” on Apple Podcasts, Spotify, or wherever you get your podcasts. Get alerts about each new episode by signing up for the “First Opinion Podcast” newsletter. And don’t forget to sign up for the First Opinion newsletter, delivered every Sunday.

Torie Bosch: Jake Eberts did not die of dysentery. But he did catch it for science. How much would you have to be paid to risk a bout with a disease that most Americans associate with the Oregon Trail?

Read the rest…

  • ✇TechCrunch
  • The browser wars are back, and this time they’re powered by AI Theresa Loconsolo
    The browser wars are heating up again, this time with AI in the driver’s seat.  OpenAI just launched Atlas, a ChatGPT-powered browser that lets users surf the web using natural language, and even includes an “agent mode” that can complete tasks autonomously. It’s one of the biggest browser launches in recent memory, but it’s debuting […]
     
  • 🏷️
  • ↗️

The browser wars are back, and this time they’re powered by AI

✇TechCrunch
By: Theresa Loconsolo
25 October 2025 at 03:00
The browser wars are heating up again, this time with AI in the driver’s seat.  OpenAI just launched Atlas, a ChatGPT-powered browser that lets users surf the web using natural language, and even includes an “agent mode” that can complete tasks autonomously. It’s one of the biggest browser launches in recent memory, but it’s debuting […]
Received — 24 October 2025 ⏭ Favourites

Bias by Design? How Data Practices Shape Fairness in AI Healthcare Systems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Anna Arias-Duart · Maria Eugenia Cardello · Atia Cort\'es
24 October 2025 at 12:00
arXiv:2510.20332v1 Announce Type: new Abstract: Artificial intelligence (AI) holds great promise for transforming healthcare. However, despite significant advances, the integration of AI solutions into real-world clinical practice remains limited. A major barrier is the quality and fairness of training data, which is often compromised by biased data collection practices. This paper draws on insights from the AI4HealthyAging project, part of Spain's national R&D initiative, where our task was to detect biases during clinical data collection. We identify several types of bias across multiple use cases, including historical, representation, and measurement biases. These biases manifest in variables such as sex, gender, age, habitat, socioeconomic status, equipment, and labeling. We conclude with practical recommendations for improving the fairness and robustness of clinical problem design and data collection. We hope that our findings and experience contribute to guiding future projects in the development of fairer AI systems in healthcare.

FLORA: Unsupervised Knowledge Graph Alignment by Fuzzy Logic

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yiwen Peng (IP Paris) · Thomas Bonald (IP Paris) · Fabian M. Suchanek (IP Paris)
24 October 2025 at 12:00
arXiv:2510.20467v1 Announce Type: new Abstract: Knowledge graph alignment is the task of matching equivalent entities (that is, instances and classes) and relations across two knowledge graphs. Most existing methods focus on pure entity-level alignment, computing the similarity of entities in some embedding space. They lack interpretable reasoning and need training data to work. In this paper, we propose FLORA, a simple yet effective method that (1) is unsupervised, i.e., does not require training data, (2) provides a holistic alignment for entities and relations iteratively, (3) is based on fuzzy logic and thus delivers interpretable results, (4) provably converges, (5) allows dangling entities, i.e., entities without a counterpart in the other KG, and (6) achieves state-of-the-art results on major benchmarks.

Lost in Translation: Policymakers are not really listening to Citizen Concerns about AI

✇cs.AI, q-bio.NC updates on arXiv.org
By: Susan Ariel Aaronson · Michael Moreno
24 October 2025 at 12:00
arXiv:2510.20568v1 Announce Type: new Abstract: The worlds people have strong opinions about artificial intelligence (AI), and they want policymakers to listen. Governments are inviting public comment on AI, but as they translate input into policy, much of what citizens say is lost. Policymakers are missing a critical opportunity to build trust in AI and its governance. This paper compares three countries, Australia, Colombia, and the United States, that invited citizens to comment on AI risks and policies. Using a landscape analysis, the authors examined how each government solicited feedback and whether that input shaped governance. Yet in none of the three cases did citizens and policymakers establish a meaningful dialogue. Governments did little to attract diverse voices or publicize calls for comment, leaving most citizens unaware or unprepared to respond. In each nation, fewer than one percent of the population participated. Moreover, officials showed limited responsiveness to the feedback they received, failing to create an effective feedback loop. The study finds a persistent gap between the promise and practice of participatory AI governance. The authors conclude that current approaches are unlikely to build trust or legitimacy in AI because policymakers are not adequately listening or responding to public concerns. They offer eight recommendations: promote AI literacy; monitor public feedback; broaden outreach; hold regular online forums; use innovative engagement methods; include underrepresented groups; respond publicly to input; and make participation easier.
  • ✇cs.AI, q-bio.NC updates on arXiv.org
  • Fluidity Index: Next-Generation Super-intelligence Benchmarks Eric Ngoiya · Tianshu Bao
    arXiv:2510.20636v1 Announce Type: new Abstract: This paper introduces the Fluidity Index (FI) to quantify model adaptability in dynamic, scaling environments. The benchmark evaluates response accuracy based on deviations in initial, current, and future environment states, assessing context switching and continuity. We distinguish between closed-ended and open-ended benchmarks, prioritizing closed-loop open-ended real-world benchmarks to test adaptability. The approach measures a model's ability
     
  • 🏷️
  • ↗️

Fluidity Index: Next-Generation Super-intelligence Benchmarks

✇cs.AI, q-bio.NC updates on arXiv.org
By: Eric Ngoiya · Tianshu Bao
24 October 2025 at 12:00
arXiv:2510.20636v1 Announce Type: new Abstract: This paper introduces the Fluidity Index (FI) to quantify model adaptability in dynamic, scaling environments. The benchmark evaluates response accuracy based on deviations in initial, current, and future environment states, assessing context switching and continuity. We distinguish between closed-ended and open-ended benchmarks, prioritizing closed-loop open-ended real-world benchmarks to test adaptability. The approach measures a model's ability to understand, predict, and adjust to state changes in scaling environments. A truly super-intelligent model should exhibit at least second-order adaptability, enabling self-sustained computation through digital replenishment for optimal fluidity.

MolBridge: Atom-Level Joint Graph Refinement for Robust Drug-Drug Interaction Event Prediction

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xuan Lin · Aocheng Ding · Tengfei Ma · Hua Liang · Zhe Quan
24 October 2025 at 12:00
arXiv:2510.20448v1 Announce Type: cross Abstract: Drug combinations offer therapeutic benefits but also carry the risk of adverse drug-drug interactions (DDIs), especially under complex molecular structures. Accurate DDI event prediction requires capturing fine-grained inter-drug relationships, which are critical for modeling metabolic mechanisms such as enzyme-mediated competition. However, existing approaches typically rely on isolated drug representations and fail to explicitly model atom-level cross-molecular interactions, limiting their effectiveness across diverse molecular complexities and DDI type distributions. To address these limitations, we propose MolBridge, a novel atom-level joint graph refinement framework for robust DDI event prediction. MolBridge constructs a joint graph that integrates atomic structures of drug pairs, enabling direct modeling of inter-drug associations. A central challenge in such joint graph settings is the potential loss of information caused by over-smoothing when modeling long-range atomic dependencies. To overcome this, we introduce a structure consistency module that iteratively refines node features while preserving the global structural context. This joint design allows MolBridge to effectively learn both local and global interaction outperforms state-of-the-art baselines, achieving superior performance across long-tail and inductive scenarios. patterns, yielding robust representations across both frequent and rare DDI types. Extensive experiments on two benchmark datasets show that MolBridge consistently. These results demonstrate the advantages of fine-grained graph refinement in improving the accuracy, robustness, and mechanistic interpretability of DDI event prediction.This work contributes to Web Mining and Content Analysis by developing graph-based methods for mining and analyzing drug-drug interaction networks.

User Perceptions of Privacy and Helpfulness in LLM Responses to Privacy-Sensitive Scenarios

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xiaoyuan Wu · Roshni Kaushik · Wenkai Li · Lujo Bauer · Koichi Onoue
24 October 2025 at 12:00
arXiv:2510.20721v1 Announce Type: cross Abstract: Large language models (LLMs) have seen rapid adoption for tasks such as drafting emails, summarizing meetings, and answering health questions. In such uses, users may need to share private information (e.g., health records, contact details). To evaluate LLMs' ability to identify and redact such private information, prior work developed benchmarks (e.g., ConfAIde, PrivacyLens) with real-life scenarios. Using these benchmarks, researchers have found that LLMs sometimes fail to keep secrets private when responding to complex tasks (e.g., leaking employee salaries in meeting summaries). However, these evaluations rely on LLMs (proxy LLMs) to gauge compliance with privacy norms, overlooking real users' perceptions. Moreover, prior work primarily focused on the privacy-preservation quality of responses, without investigating nuanced differences in helpfulness. To understand how users perceive the privacy-preservation quality and helpfulness of LLM responses to privacy-sensitive scenarios, we conducted a user study with 94 participants using 90 scenarios from PrivacyLens. We found that, when evaluating identical responses to the same scenario, users showed low agreement with each other on the privacy-preservation quality and helpfulness of the LLM response. Further, we found high agreement among five proxy LLMs, while each individual LLM had low correlation with users' evaluations. These results indicate that the privacy and helpfulness of LLM responses are often specific to individuals, and proxy LLMs are poor estimates of how real users would perceive these responses in privacy-sensitive scenarios. Our results suggest the need to conduct user-centered studies on measuring LLMs' ability to help users while preserving privacy. Additionally, future research could investigate ways to improve the alignment between proxy LLMs and users for better estimation of users' perceived privacy and utility.

Automated Extraction of Fluoropyrimidine Treatment and Treatment-Related Toxicities from Clinical Notes Using Natural Language Processing

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xizhi Wu · Madeline S. Kreider · Philip E. Empey · Chenyu Li · Yanshan Wang
24 October 2025 at 12:00
arXiv:2510.20727v1 Announce Type: cross Abstract: Objective: Fluoropyrimidines are widely prescribed for colorectal and breast cancers, but are associated with toxicities such as hand-foot syndrome and cardiotoxicity. Since toxicity documentation is often embedded in clinical notes, we aimed to develop and evaluate natural language processing (NLP) methods to extract treatment and toxicity information. Materials and Methods: We constructed a gold-standard dataset of 236 clinical notes from 204,165 adult oncology patients. Domain experts annotated categories related to treatment regimens and toxicities. We developed rule-based, machine learning-based (Random Forest, Support Vector Machine [SVM], Logistic Regression [LR]), deep learning-based (BERT, ClinicalBERT), and large language models (LLM)-based NLP approaches (zero-shot and error-analysis prompting). Models used an 80:20 train-test split. Results: Sufficient data existed to train and evaluate 5 annotated categories. Error-analysis prompting achieved optimal precision, recall, and F1 scores (F1=1.000) for treatment and toxicities extraction, whereas zero-shot prompting reached F1=1.000 for treatment and F1=0.876 for toxicities extraction.LR and SVM ranked second for toxicities (F1=0.937). Deep learning underperformed, with BERT (F1=0.873 treatment; F1= 0.839 toxicities) and ClinicalBERT (F1=0.873 treatment; F1 = 0.886 toxicities). Rule-based methods served as our baseline with F1 scores of 0.857 in treatment and 0.858 in toxicities. Discussion: LMM-based approaches outperformed all others, followed by machine learning methods. Machine and deep learning approaches were limited by small training data and showed limited generalizability, particularly for rare categories. Conclusion: LLM-based NLP most effectively extracted fluoropyrimidine treatment and toxicity information from clinical notes, and has strong potential to support oncology research and pharmacovigilance.

FieldGen: From Teleoperated Pre-Manipulation Trajectories to Field-Guided Data Generation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Wenhao Wang · Kehe Ye · Xinyu Zhou · Tianxing Chen · Cao Min · Qiaoming Zhu · Xiaokang Yang · Yongjian Shen · Yang Yang · Maoqing Yao · Yao Mu
24 October 2025 at 12:00
arXiv:2510.20774v1 Announce Type: cross Abstract: Large-scale and diverse datasets are vital for training robust robotic manipulation policies, yet existing data collection methods struggle to balance scale, diversity, and quality. Simulation offers scalability but suffers from sim-to-real gaps, while teleoperation yields high-quality demonstrations with limited diversity and high labor cost. We introduce FieldGen, a field-guided data generation framework that enables scalable, diverse, and high-quality real-world data collection with minimal human supervision. FieldGen decomposes manipulation into two stages: a pre-manipulation phase, allowing trajectory diversity, and a fine manipulation phase requiring expert precision. Human demonstrations capture key contact and pose information, after which an attraction field automatically generates diverse trajectories converging to successful configurations. This decoupled design combines scalable trajectory diversity with precise supervision. Moreover, FieldGen-Reward augments generated data with reward annotations to further enhance policy learning. Experiments demonstrate that policies trained with FieldGen achieve higher success rates and improved stability compared to teleoperation-based baselines, while significantly reducing human effort in long-term real-world data collection. Webpage is available at https://fieldgen.github.io/.

Annotation Guidelines-Based Knowledge Augmentation: Towards Enhancing Large Language Models for Educational Text Classification

✇cs.AI, q-bio.NC updates on arXiv.org
By: Shiqi Liu · Sannyuya Liu · Lele Sha · Zijie Zeng · Dragan Gasevic · Zhi Liu
24 October 2025 at 12:00
arXiv:2406.00954v2 Announce Type: replace-cross Abstract: Various machine learning approaches have gained significant popularity for the automated classification of educational text to identify indicators of learning engagement -- i.e. learning engagement classification (LEC). LEC can offer comprehensive insights into human learning processes, attracting significant interest from diverse research communities, including Natural Language Processing (NLP), Learning Analytics, and Educational Data Mining. Recently, Large Language Models (LLMs), such as ChatGPT, have demonstrated remarkable performance in various NLP tasks. However, their comprehensive evaluation and improvement approaches in LEC tasks have not been thoroughly investigated. In this study, we propose the Annotation Guidelines-based Knowledge Augmentation (AGKA) approach to improve LLMs. AGKA employs GPT 4.0 to retrieve label definition knowledge from annotation guidelines, and then applies the random under-sampler to select a few typical examples. Subsequently, we conduct a systematic evaluation benchmark of LEC, which includes six LEC datasets covering behavior classification (question and urgency level), emotion classification (binary and epistemic emotion), and cognition classification (opinion and cognitive presence). The study results demonstrate that AGKA can enhance non-fine-tuned LLMs, particularly GPT 4.0 and Llama 3 70B. GPT 4.0 with AGKA few-shot outperforms full-shot fine-tuned models such as BERT and RoBERTa on simple binary classification datasets. However, GPT 4.0 lags in multi-class tasks that require a deep understanding of complex semantic information. Notably, Llama 3 70B with AGKA is a promising combination based on open-source LLM, because its performance is on par with closed-source GPT 4.0 with AGKA. In addition, LLMs struggle to distinguish between labels with similar names in multi-class classification.

Serving LLMs in HPC Clusters: A Comparative Study of Qualcomm Cloud AI 100 Ultra and NVIDIA Data Center GPUs

✇cs.AI, q-bio.NC updates on arXiv.org
By: Mohammad Firas Sada, John J. Graham, Elham E Khoda, Mahidhar Tatineni, Dmitry Mishin, Rajesh K. Gupta, Rick Wagner, Larry Smarr, Thomas A. DeFanti, Frank W\"urthwein
24 October 2025 at 12:00
arXiv:2507.00418v2 Announce Type: replace-cross Abstract: This study presents a benchmarking analysis of the Qualcomm Cloud AI 100 Ultra (QAic) accelerator for large language model (LLM) inference, evaluating its energy efficiency (throughput per watt), performance, and hardware scalability against NVIDIA A100 GPUs (in 4x and 8x configurations) within the National Research Platform (NRP) ecosystem. A total of 12 open-source LLMs, ranging from 124 million to 70 billion parameters, are served using the vLLM framework. Our analysis reveals that QAic achieves competitive energy efficiency with advantages on specific models while enabling more granular hardware allocation: some 70B models operate on as few as 1 QAic card versus 8 A100 GPUs required, with 20x lower power consumption (148W vs 2,983W). For smaller models, single QAic devices achieve up to 35x lower power consumption compared to our 4-GPU A100 configuration (36W vs 1,246W). The findings offer insights into the potential of the Qualcomm Cloud AI 100 Ultra for energy-constrained and resource-efficient HPC deployments within the National Research Platform (NRP).

Position: The Current AI Conference Model is Unsustainable! Diagnosing the Crisis of Centralized AI Conference

✇cs.AI, q-bio.NC updates on arXiv.org
By: Nuo Chen · Moming Duan · Andre Huikai Lin · Qian Wang · Jiaying Wu · Bingsheng He
24 October 2025 at 12:00
arXiv:2508.04586v4 Announce Type: replace-cross Abstract: Artificial Intelligence (AI) conferences are essential for advancing research, sharing knowledge, and fostering academic community. However, their rapid expansion has rendered the centralized conference model increasingly unsustainable. This paper offers a data-driven diagnosis of a structural crisis that threatens the foundational goals of scientific dissemination, equity, and community well-being. We identify four key areas of strain: (1) scientifically, with per-author publication rates more than doubling over the past decade to over 4.5 papers annually; (2) environmentally, with the carbon footprint of a single conference exceeding the daily emissions of its host city; (3) psychologically, with 71% of online community discourse reflecting negative sentiment and 35% referencing mental health concerns; and (4) logistically, with attendance at top conferences such as NeurIPS 2024 beginning to outpace venue capacity. These pressures point to a system that is misaligned with its core mission. In response, we propose the Community-Federated Conference (CFC) model, which separates peer review, presentation, and networking into globally coordinated but locally organized components, offering a more sustainable, inclusive, and resilient path forward for AI research.

VaultGemma: A Differentially Private Gemma Model

✇cs.AI, q-bio.NC updates on arXiv.org
By: Amer Sinha · Thomas Mesnard · Ryan McKenna · Daogao Liu · Christopher A. Choquette-Choo · Yangsibo Huang · Da Yu · George Kaissis · Zachary Charles · Ruibo Liu · Lynn Chua · Pritish Kamath · Pasin Manurangsi · Steve He · Chiyuan Zhang · Badih Ghazi · Borja De Balle Pigem · Prem Eruvbetine · Tris Warkentin · Armand Joulin · Ravi Kumar
24 October 2025 at 12:00
arXiv:2510.15001v2 Announce Type: replace-cross Abstract: We introduce VaultGemma 1B, a 1 billion parameter model within the Gemma family, fully trained with differential privacy. Pretrained on the identical data mixture used for the Gemma 2 series, VaultGemma 1B represents a significant step forward in privacy-preserving large language models. We openly release this model to the community
Received — 23 October 2025 ⏭ Favourites

A Multi-faceted Analysis of Cognitive Abilities: Evaluating Prompt Methods with Large Language Models on the CONSORT Checklist

✇cs.AI, q-bio.NC updates on arXiv.org
By: Sohyeon Jeon · Hyung-Chul Lee
23 October 2025 at 12:00
arXiv:2510.19139v1 Announce Type: new Abstract: Despite the rapid expansion of Large Language Models (LLMs) in healthcare, the ability of these systems to assess clinical trial reporting according to CONSORT standards remains unclear, particularly with respect to their cognitive and reasoning strategies. This study applies a behavioral and metacognitive analytic approach with expert-validated data, systematically comparing two representative LLMs under three prompt conditions. Clear differences emerged in how the models approached various CONSORT items, and prompt types, including shifts in reasoning style, explicit uncertainty, and alternative interpretations shaped response patterns. Our results highlight the current limitations of these systems in clinical compliance automation and underscore the importance of understanding their cognitive adaptations and strategic behavior in developing more explainable and reliable medical AI.

MSC-Bench: A Rigorous Benchmark for Multi-Server Tool Orchestration

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jia-Kai Dong · I-Wei Huang · Chun-Tin Wu · Yi-Tien Tsai
23 October 2025 at 12:00
arXiv:2510.19423v1 Announce Type: new Abstract: We introduce MSC-Bench, a large-scale benchmark for evaluating multi-hop, end-to-end tool orchestration by LLM agents in a hierarchical Model-Context Protocol (MCP) ecosystem. Existing benchmarks often evaluate tools in isolation, ignoring challenges such as functional overlap and cross-server orchestration, leading to overly optimistic assessments. MSC-Bench addresses these gaps by constructing ground truth through 'equal function sets', allowing objective metrics such as F1 score and reducing the dependency on LLM-as-a-judge evaluation. Organized as a five-level curriculum, it systematically tests agent capabilities from single-tool orchestration to complex cross-server planning, and robustness to out-of-scope requests. Experiments reveal that rigid hierarchies can hinder performance without co-designed strategies, and even state-of-the-art agents exhibit systemic weaknesses in robustness. MSC-Bench provides a diagnostic framework to expose these limitations and guide the development of more capable and efficient tool-using agents. The benchmark and resources are publicly available at https://github.com/snooow1029/MSC_Bench.
❌