arXiv:2510.25813v1 Announce Type: new
Abstract: We present a novel framework for Industry 5.0 that simplifies the deployment of AI models on edge devices in various industrial settings. The design reduces latency and avoids external data transfer by enabling local inference and real-time processing. Our implementation is agent-based, which means that individual agents, whether human, algorithmic, or collaborative, are responsible for well-defined tasks, enabling flexibility and simplifying integration. Moreover, our framework supports modular integration and maintains low resource requirements. Preliminary evaluations concerning the food industry in real scenarios indicate improved deployment time and system adaptability performance. The source code is publicly available at https://github.com/AI-REDGIO-5-0/ci-component.
arXiv:2510.25819v1 Announce Type: cross
Abstract: The rapid rise of AI agents presents urgent challenges in authentication, authorization, and identity management. Current agent-centric protocols (like MCP) highlight the demand for clarified best practices in authentication and authorization. Looking ahead, ambitions for highly autonomous agents raise complex long-term questions regarding scalable access control, agent-centric identities, AI workload differentiation, and delegated authority. This OpenID Foundation whitepaper is for stakeholders at the intersection of AI agents and access management. It outlines the resources already available for securing today's agents and presents a strategic agenda to address the foundational authentication, authorization, and identity problems pivotal for tomorrow's widespread autonomous systems.
arXiv:2510.25992v1 Announce Type: cross
Abstract: Large Language Models (LLMs) often struggle with problems that require multi-step reasoning. For small-scale open-source models, Reinforcement Learning with Verifiable Rewards (RLVR) fails when correct solutions are rarely sampled even after many attempts, while Supervised Fine-Tuning (SFT) tends to overfit long demonstrations through rigid token-by-token imitation. To address this gap, we propose Supervised Reinforcement Learning (SRL), a framework that reformulates problem solving as generating a sequence of logical "actions". SRL trains the model to generate an internal reasoning monologue before committing to each action. It provides smoother rewards based on the similarity between the model's actions and expert actions extracted from the SFT dataset in a step-wise manner. This supervision offers richer learning signals even when all rollouts are incorrect, while encouraging flexible reasoning guided by expert demonstrations. As a result, SRL enables small models to learn challenging problems previously unlearnable by SFT or RLVR. Moreover, initializing training with SRL before refining with RLVR yields the strongest overall performance. Beyond reasoning benchmarks, SRL generalizes effectively to agentic software engineering tasks, establishing it as a robust and versatile training framework for reasoning-oriented LLMs.
arXiv:2510.26007v1 Announce Type: cross
Abstract: The development of Artificial Intelligence (AI), including AI in Science (AIS), should be done following the principles of responsible AI. Progress in responsible AI is often quantified through evaluation metrics, yet there has been less work on assessing the robustness and reliability of the metrics themselves. We reflect on prior work that examines the robustness of fairness metrics for recommender systems as a type of AI application and summarise their key takeaways into a set of non-exhaustive guidelines for developing reliable metrics of responsible AI. Our guidelines apply to a broad spectrum of AI applications, including AIS.
arXiv:2510.23595v3 Announce Type: replace
Abstract: Reinforcement Learning (RL) has demonstrated significant potential in enhancing the reasoning capabilities of large language models (LLMs). However, the success of RL for LLMs heavily relies on human-curated datasets and verifiable rewards, which limit their scalability and generality. Recent Self-Play RL methods, inspired by the success of the paradigm in games and Go, aim to enhance LLM reasoning capabilities without human-annotated data. However, their methods primarily depend on a grounded environment for feedback (e.g., a Python interpreter or a game engine); extending them to general domains remains challenging. To address these challenges, we propose Multi-Agent Evolve (MAE), a framework that enables LLMs to self-evolve in solving diverse tasks, including mathematics, reasoning, and general knowledge Q&A. The core design of MAE is based on a triplet of interacting agents (Proposer, Solver, Judge) that are instantiated from a single LLM, and applies reinforcement learning to optimize their behaviors. The Proposer generates questions, the Solver attempts solutions, and the Judge evaluates both while co-evolving. Experiments on Qwen2.5-3B-Instruct demonstrate that MAE achieves an average improvement of 4.54% on multiple benchmarks. These results highlight MAE as a scalable, data-efficient method for enhancing the general reasoning abilities of LLMs with minimal reliance on human-curated supervision.
arXiv:2406.05948v4 Announce Type: replace-cross
Abstract: Large Language Models (LLMs), especially those accessed via APIs, have demonstrated impressive capabilities across various domains. However, users without technical expertise often turn to (untrustworthy) third-party services, such as prompt engineering, to enhance their LLM experience, creating vulnerabilities to adversarial threats like backdoor attacks. Backdoor-compromised LLMs generate malicious outputs to users when inputs contain specific "triggers" set by attackers. Traditional defense strategies, originally designed for small-scale models, are impractical for API-accessible LLMs due to limited model access, high computational costs, and data requirements. To address these limitations, we propose Chain-of-Scrutiny (CoS) which leverages LLMs' unique reasoning abilities to mitigate backdoor attacks. It guides the LLM to generate reasoning steps for a given input and scrutinizes for consistency with the final output -- any inconsistencies indicating a potential attack. It is well-suited for the popular API-only LLM deployments, enabling detection at minimal cost and with little data. User-friendly and driven by natural language, it allows non-experts to perform the defense independently while maintaining transparency. We validate the effectiveness of CoS through extensive experiments on various tasks and LLMs, with results showing greater benefits for more powerful LLMs.
arXiv:2410.14879v4 Announce Type: replace-cross
Abstract: Passive tracking methods, such as phone and wearable sensing, have become dominant in monitoring human behaviors in modern ubiquitous computing studies. While there have been significant advances in machine-learning approaches to translate periods of raw sensor data to model momentary behaviors, (e.g., physical activity recognition), there still remains a significant gap in the translation of these sensing streams into meaningful, high-level, context-aware insights that are required for various applications (e.g., summarizing an individual's daily routine). To bridge this gap, experts often need to employ a context-driven sensemaking process in real-world studies to derive insights. This process often requires manual effort and can be challenging even for experienced researchers due to the complexity of human behaviors.
We conducted three rounds of user studies with 21 experts to explore solutions to address challenges with sensemaking. We follow a human-centered design process to identify needs and design, iterate, build, and evaluate Vital Insight (VI), a novel, LLM-assisted, prototype system to enable human-in-the-loop inference (sensemaking) and visualizations of multi-modal passive sensing data from smartphones and wearables. Using the prototype as a technology probe, we observe experts' interactions with it and develop an expert sensemaking model that explains how experts move between direct data representations and AI-supported inferences to explore, question, and validate insights. Through this iterative process, we also synthesize and discuss a list of design implications for the design of future AI-augmented visualization systems to better assist experts' sensemaking processes in multi-modal health sensing data.
arXiv:2510.04226v4 Announce Type: replace-cross
Abstract: Large language models (LLMs) tend to generate lexically, semantically, and stylistically homogenous texts. This poses a risk of knowledge collapse, where homogenous LLMs mediate a shrinking in the range of accessible information over time. Existing works on homogenization are limited by a focus on closed-ended multiple-choice setups or fuzzy semantic features, and do not look at trends across time and cultural contexts. To overcome this, we present a new methodology to measure epistemic diversity, i.e., variation in real-world claims in LLM outputs, which we use to perform a broad empirical study of LLM knowledge collapse. We test 27 LLMs, 155 topics covering 12 countries, and 200 prompt variations sourced from real user chats. For the topics in our study, we show that while newer models tend to generate more diverse claims, nearly all models are less epistemically diverse than a basic web search. We find that model size has a negative impact on epistemic diversity, while retrieval-augmented generation (RAG) has a positive impact, though the improvement from RAG varies by the cultural context. Finally, compared to a traditional knowledge source (Wikipedia), we find that country-specific claims reflect the English language more than the local one, highlighting a gap in epistemic representation
arXiv:2510.23639v2 Announce Type: replace-cross
Abstract: This paper introduces an innovative Electronic Health Record (EHR) foundation model that integrates Polygenic Risk Scores (PRS) as a foundational data modality, moving beyond traditional EHR-only approaches to build more holistic health profiles. Leveraging the extensive and diverse data from the All of Us (AoU) Research Program, this multimodal framework aims to learn complex relationships between clinical data and genetic predispositions. The methodology extends advancements in generative AI to the EHR foundation model space, enhancing predictive capabilities and interpretability. Evaluation on AoU data demonstrates the model's predictive value for the onset of various conditions, particularly Type 2 Diabetes (T2D), and illustrates the interplay between PRS and EHR data. The work also explores transfer learning for custom classification tasks, showcasing the architecture's versatility and efficiency. This approach is pivotal for unlocking new insights into disease prediction, proactive health management, risk stratification, and personalized treatment strategies, laying the groundwork for more personalized, equitable, and actionable real-world evidence generation in healthcare.
Front Immunol. 2025 Oct 14;16:1668630. doi: 10.3389/fimmu.2025.1668630. eCollection 2025.
ABSTRACT
AIMS AND OBJECTIVES: This study aimed to identify immunologically relevant transcriptomic and proteomic biomarkers in hepatocellular carcinoma (HCC) and to characterize their B-cell epitopes for potential integration into nanomaterial-based biosensors and immunomodulatory platforms for early diagnosis and targeted therapy.
METHODS: We conducted a comprehensive multi-omics analysis by integrating transcriptomic (TCGA-LIHC) and proteomic data to identify differentially expressed genes (DEGs) in HCC. Protein-protein interaction networks and pathway enrichment were used to prioritize hub genes. Five candidate biomarkers, RFC2, HSP90AB1, YWHAZ, CYP2E1, and ADH4, were selected for qRT-PCR and serum ELISA validation in clinical cohorts comprising 85 HCC patients and 50 healthy controls. B-cell epitope prediction was performed using BepiPred 2.0 and validated through synthetic peptide-based ELISA in the same cohort to assess immunoreactivity. Diagnostic performance was evaluated using ROC curve analysis.
RESULTS: RFC2, HSP90AB1, and YWHAZ were significantly upregulated (|log2FC|>0.2) and showed high serological expression, whereas CYP2E1 and ADH4 were consistently downregulated. Predicted B-cell epitopes from RFC2, HSP90AB1, and YWHAZ exhibited strong immunoreactivity (AUC>0.84), indicating their diagnostic potential. Enrichment analysis revealed that upregulated DEGs were involved in cell cycle and mitotic progression, while downregulated genes were linked to immune suppression and metabolic dysfunction. These validated immunogenic epitopes offer promising anchors for nanomaterial-functionalized biosensors, such as gold nanoparticle-conjugated ELISA, graphene-based electrochemical platforms, and peptide-coated quantum dots, for ultrasensitive and multiplexed HCC detection.
CONCLUSION: By integrating transcriptomic and proteomic screening with epitope-level validation, we identified a novel panel of immunogenic biomarkers suitable for nanomaterial-enabled diagnostics in HCC. These findings support the translational potential of peptide-nano scaffold conjugates in developing minimally invasive, immune-responsive biosensing and therapeutic tools tailored for early-stage liver cancer management.
Four years is a lifetime when it comes to artificial intelligence. Since the first edition of this study was published in 2021, AI’s capabilities have been advancing at speed, and the advances have not slowed since generative AI’s breakthrough. For example, multimodality— the ability to process information not only as text but also as audio, video, and other unstructured formats—is becoming a common feature of AI models. AI’s capacity to reason and act autonomously has also grown, and organizations are now starting to work with AI agents that can do just that.
Amid all the change, there remains a constant: the quality of an AI model’s outputs is only ever as good as the data that feeds it. Data management technologies and practices have also been advancing, but the second edition of this study suggests that most organizations are not leveraging those fast enough to keep up with AI’s development. As a result of that and other hindrances, relatively few organizations are delivering the desired business results from their AI strategy. No more than 2% of senior executives we surveyed rate their organizations highly in terms of delivering results from AI.
To determine the extent to which organizational data performance has improved as generative AI and other AI advances have taken hold, MIT Technology Review Insights surveyed 800 senior data and technology executives. We also conducted in-depth interviews with 15 technology and business leaders.
Key findings from the report include the following:
• Few data teams are keeping pace with AI. Organizations are doing no better today at delivering on data strategy than in pre-generative AI days. Among those surveyed in 2025, 12% are self-assessed data “high achievers” compared with 13% in 2021. Shortages of skilled talent remain a constraint, but teams also struggle with accessing fresh data, tracing lineage, and dealing with security complexity—important requirements for AI success.
• Partly as a result, AI is not fully firing yet. There are even fewer “high achievers” when it comes to AI. Just 2% of respondents rate their organizations’ AI performance highly today in terms of delivering measurable business results. In fact, most are still struggling to scale generative AI. While two thirds have deployed it, only 7% have done so widely.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. AI tools that may have been used were limited to secondary production processes that passed thorough human review.
This multi-omic longitudinal analysis of the healthy human peripheral immune system constructs the Human Immune Health Atlas and assembles data on immune cell composition and state changes with age, including responses to cytomegalovirus infection and influenza vaccination.
Diagnostics (Basel). 2025 Oct 14;15(20):2586. doi: 10.3390/diagnostics15202586.
ABSTRACT
The integration of multi-omics technologies is transforming the landscape of cancer management, offering unprecedented insights into tumor biology, early diagnosis, and personalized therapy. This review provides a comprehensive overview of the current state of omics approaches, with a particular focus on the application of genomics, NMR-based metabolomics, and radiomics in non-small cell lung cancer (NSCLC). Genomics currently represents one of the most established omics technologies in oncology, as it enables the identification of genetic alterations that drive tumor initiation, progression, and therapeutic response. Interestingly, genomic analyses have revealed that many tumors harbor mutations in genes encoding metabolic enzymes, thus establishing a tight connection between genomics and tumor metabolism. In parallel, metabolomics profiling-by capturing the metabolic phenotype of tumors-has, in recent years, identified specific biomarkers associated with tumor burden, progression, and prognosis. Such findings have catalyzed growing interest in metabolomics as a complementary approach to better characterize cancer biology and discover novel diagnostic and therapeutic targets. Moreover, radiomics, through the extraction of quantitative features from standard imaging modalities, captures tumor heterogeneity and contributes predictive information on tumor biology, treatment response, and clinical outcomes. As a non-invasive and widely available technique, radiomics has the potential to support longitudinal monitoring and individualized treatment planning. Both metabolomics and radiomics, when integrated with genomic data, could support a more comprehensive understanding of NSCLC and pave the way for the development of non-invasive, predictive models and personalized therapeutic strategies. In addition, we explore the specific contributions of these technologies in enhancing clinical decision-making for lung cancer patients, with particular attention to their potential in early diagnosis, treatment selection, and real-time monitoring.
BACKGROUND AND OBJECTIVE: Lung adenocarcinoma (LUAD), as the main subtype of non-small cell lung cancer (NSCLC), faces clinical challenges including molecular heterogeneity, late diagnosis, and aggressive growth, leading to a low 5-year survival rate. Biomarkers are critical for early detection, accurate differentiation of benign/malignant lesions, and guiding personalized treatment strategies. Proteomic technologies using liquid biopsy show potential by analyzing protein changes and post-translational modifications (PTMs) to identify novel biomarkers and unravel cancer mechanisms. This review examines proteomic advances in LUAD, compares platform strengths, lists validated protein markers, and discusses challenges like specificity and regulations. It aims to develop a precision medicine framework by integrating multi-omics data for improved diagnosis and treatment.
METHODS: This study conducted a literature review by searching the PubMed and Web of Science databases for original articles written in English from 2002 to 2025, using the keywords "lung adenocarcinoma" OR "LUAD" AND "biomarkers" AND "proteomics" OR "SomaScan" OR "spatial proteomics" to identify the latest research findings in the field of proteomics technology and LUAD biomarkers. The included studies mainly focused on the current landscape of biomarkers in the diagnosis, treatment, and prognosis of LUAD.
KEY CONTENT AND FINDINGS: This review discusses high-throughput methods for comprehensive protein profiling in accessible biospecimens (tissues, blood, urine) to identify biomarkers for LUAD. We systematically evaluate emerging proteomic strategies, including mass spectrometry (MS), proximity extension assays (PEAs), spatial proteomics techniques, and SomaScan platforms-coupled with innovative computational frameworks have revolutionized biomarkers discovery and their translational potential in developing precision diagnostics and targeted therapies. Additionally, the review addresses challenges in integrating proteomics with genomics, transcriptomics, and metabolomics, offering new methodologies and expanding research in life sciences. As technological advancements continue, it is anticipated that more potential biomarkers will be conducted to validate the broader application in LUAD treatment, addressing early-stage disease complexities and aiding in selecting more effective treatment strategies.
CONCLUSIONS: By synthesizing cutting-edge evidence on proteome-driven LUAD biomarkers, this review elucidates actionable strategies to refine early detection protocols and mechanism-informed personalized treatment frameworks, directly advancing precision oncology initiatives for this prevalent malignancy through biomarker-guided clinical decision-making and multi-omics integration.
Want to stay on top of the science and politics driving biotech today? Sign up to get our biotech newsletter in your inbox.
Good morning. It seems everyone I know has been getting sick lately — hope you are all taking care of yourselves! Onto the news today.
BridgeBio notches another Phase 3 win
BridgeBio said this morning that its investigational drug succeeded in a late-stage trial of patients with autosomal dominant hypocalcemia type 1, a rare genetic condition that causes low calcium levels in the blood.
Background: Patient recruitment remains a critical challenge in clinical research. Although the integration of electronically collected patient-entered data within clinical practices enables innovative recruitment approaches, existing methods present challenges such as increased patient burden and potential violation of autonomy. A more nuanced approach involves identifying patient attributes associated with higher propensity for research participation, enabling research teams to efficiently prioritize outreach efforts. Objective: This study aims to (1) develop patient-reported questions reflecting perceptions about research participation and (2) determine whether patient responses are predictive of interest in joining a precision medicine registry. Methods: This mixed methods study used an exploratory sequential design in 2 phases. Phase 1 involved cognitive interviews with 32 patients recruited through the Cleveland Clinic Healthcare Partners program to develop “research perception” questions. Participants evaluated 9 candidate questions that were based on a literature review of research participation factors. Three questions were selected for implementation. Phase 2 was a cross-sectional cohort study incorporating these 3 questions into routine electronic questionnaires completed by primary care patients through the patient portal. The study population included 1077 patients who completed both “research perception” and “research recruitment” questions between August 2018 and April 2019. Diagnostic accuracy was assessed using receiver operating characteristic curve analysis, and multivariable logistic regression models evaluated associations while adjusting for demographic and health factors. Results: Phase 1 revealed strong research support among participants, with 97% (31/32) agreeing that research should be part of the institution’s mission and 100% (32/32) affirming that research enhances patient care. Phase 2 included 1077 patients (mean age 48.3, SD 16.3 years; 625/1065 female, 58.68%; 661/1005 White, 65.77%), of whom 278 (25.8%) expressed interest in being contacted about the precision medicine registry. Patients expressing interest were older and had worse self-reported health, more depressive symptoms, and greater social needs. “Strongly agree” and “very important” responses to any “research perception” question were significantly associated with study interest, with adjusted odds ratios ranging from 6.36 (95% CI 2.77-14.6) to 17.6 (95% CI 5.08-61.1; P<.001). The “research perception” questions demonstrated high sensitivity (>80%) but limited specificity (24%-31%). Conclusions: Patient-reported questions assessing research participation likelihood can help identify patients more likely to enroll in clinical studies. This approach enables effective recruitment prioritization while preserving patient autonomy and reducing patient burden. High sensitivity makes these questions valuable as screening tools, although limited specificity suggests use for prioritizing rather than excluding participants. Further validation across different trial types and populations is warranted.
arXiv:2510.24161v1 Announce Type: new
Abstract: Multimodal large language models (MLLMs) have advanced vision-language reasoning and are increasingly deployed in embodied agents. However, significant limitations remain: MLLMs generalize poorly across digital-physical spaces and embodiments; vision-language-action models (VLAs) produce low-level actions yet lack robust high-level embodied reasoning; and most embodied large language models (ELLMs) are constrained to digital-space with poor generalization to the physical world. Thus, unified models that operate seamlessly across digital and physical spaces while generalizing across embodiments and tasks remain absent. We introduce the \textbf{Boundless Large Model (BLM$_1$)}, a multimodal spatial foundation model that preserves instruction following and reasoning, incorporates embodied knowledge, and supports robust cross-embodiment control. BLM$_1$ integrates three key capabilities -- \textit{cross-space transfer, cross-task learning, and cross-embodiment generalization} -- via a two-stage training paradigm. Stage I injects embodied knowledge into the MLLM through curated digital corpora while maintaining language competence. Stage II trains a policy module through an intent-bridging interface that extracts high-level semantics from the MLLM to guide control, without fine-tuning the MLLM backbone. This process is supported by a self-collected cross-embodiment demonstration suite spanning four robot embodiments and six progressively challenging tasks. Evaluations across digital and physical benchmarks show that a single BLM$_1$ instance outperforms four model families -- MLLMs, ELLMs, VLAs, and GMLMs -- achieving $\sim\!\textbf{6%}$ gains in digital tasks and $\sim\!\textbf{3%}$ in physical tasks.
arXiv:2510.24528v1 Announce Type: new
Abstract: The capability of in-context learning (ICL) enables large language models (LLMs) to perform novel tasks without parameter updates by conditioning on a few input-output examples. However, collecting high-quality examples for new or challenging tasks can be costly and labor-intensive. In this work, we propose a cost-efficient two-stage pipeline that reduces reliance on LLMs for data labeling. Our approach first leverages readily available cross-task examples to prompt an LLM and pseudo-label a small set of target task instances. We then introduce a graph-based label propagation method that spreads label information to the remaining target examples without additional LLM queries. The resulting fully pseudo-labeled dataset is used to construct in-task demonstrations for ICL. This pipeline combines the flexibility of cross-task supervision with the scalability of LLM-free propagation. Experiments across five tasks demonstrate that our method achieves strong performance while lowering labeling costs.
Login
