❌

Normal view

Received β€” 13 January 2026 ⏭ Favourites

Benchmarking Egocentric Clinical Intent Understanding Capability for Medical Multimodal Large Language Models

βœ‡cs.AI, q-bio.NC updates on arXiv.org
By: Shaonan Liu Β· Guo Yu Β· Xiaoling Luo Β· Shiyi Zheng Β· Wenting Chen Β· Jie Liu Β· Linlin Shen
13 January 2026 at 13:00
arXiv:2601.06750v1 Announce Type: cross Abstract: Medical Multimodal Large Language Models (Med-MLLMs) require egocentric clinical intent understanding for real-world deployment, yet existing benchmarks fail to evaluate this critical capability. To address these challenges, we introduce MedGaze-Bench, the first benchmark leveraging clinician gaze as a Cognitive Cursor to assess intent understanding across surgery, emergency simulation, and diagnostic interpretation. Our benchmark addresses three fundamental challenges: visual homogeneity of anatomical structures, strict temporal-causal dependencies in clinical workflows, and implicit adherence to safety protocols. We propose a Three-Dimensional Clinical Intent Framework evaluating: (1) Spatial Intent: discriminating precise targets amid visual noise, (2) Temporal Intent: inferring causal rationale through retrospective and prospective reasoning, and (3) Standard Intent: verifying protocol compliance through safety checks. Beyond accuracy metrics, we introduce Trap QA mechanisms to stress-test clinical reliability by penalizing hallucinations and cognitive sycophancy. Experiments reveal current MLLMs struggle with egocentric intent due to over-reliance on global features, leading to fabricated observations and uncritical acceptance of invalid instructions.

PulseMind: A Multi-Modal Medical Model for Real-World Clinical Diagnosis

βœ‡cs.AI, q-bio.NC updates on arXiv.org
By: Jiao Xu Β· Junwei Liu Β· Jiangwei Lao Β· Qi Zhu Β· Yunpeng Zhao Β· Congyun Jin Β· Shinan Liu Β· Zhihong Lu Β· Lihe Zhang Β· Xin Chen Β· Jian Wang Β· Ping Wang
13 January 2026 at 13:00
arXiv:2601.07344v1 Announce Type: cross Abstract: Recent advances in medical multi-modal models focus on specialized image analysis like dermatology, pathology, or radiology. However, they do not fully capture the complexity of real-world clinical diagnostics, which involve heterogeneous inputs and require ongoing contextual understanding during patient-physician interactions. To bridge this gap, we introduce PulseMind, a new family of multi-modal diagnostic models that integrates a systematically curated dataset, a comprehensive evaluation benchmark, and a tailored training framework. Specifically, we first construct a diagnostic dataset, MediScope, which comprises 98,000 real-world multi-turn consultations and 601,500 medical images, spanning over 10 major clinical departments and more than 200 sub-specialties. Then, to better reflect the requirements of real-world clinical diagnosis, we develop the PulseMind Benchmark, a multi-turn diagnostic consultation benchmark with a four-dimensional evaluation protocol comprising proactiveness, accuracy, usefulness, and language quality. Finally, we design a training framework tailored for multi-modal clinical diagnostics, centered around a core component named Comparison-based Reinforcement Policy Optimization (CRPO). Compared to absolute score rewards, CRPO uses relative preference signals from multi-dimensional com-parisons to provide stable and human-aligned training guidance. Extensive experiments demonstrate that PulseMind achieves competitive performance on both the diagnostic consultation benchmark and public medical benchmarks.
Received β€” 5 September 2025 ⏭ Favourites

Liquid Biopsy in CRC Management: Early Detection, Minimal Residual Disease, and Therapy Optimization-Clinical Evidence and Challenges

βœ‡MRD
By: Qi Liu Β· Xiaoyong Li Β· Tiejun Jin Β· Sibo Huo Β· Shuai Su Β· Nan Liu
4 September 2025 at 18:00

Diagn Cytopathol. 2025 Nov;53(11):580-591. doi: 10.1002/dc.70009. Epub 2025 Sep 4.

ABSTRACT

Colorectal cancer (CRC) is a major global health burden, ranking among the leading causes of cancer-related deaths. Despite improvements in screening and treatment, challenges such as late-stage diagnosis, high recurrence rates, and therapy resistance continue to impede optimal outcomes. Liquid biopsy, a minimally invasive technique that analyzes tumor-derived components in bodily fluids-including circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), and extracellular vesicles (EVs)-is emerging as a powerful tool to transform CRC management across the disease continuum. This review provides a comprehensive overview of liquid biopsy's current and emerging applications in CRC. We examine its role in early detection, where sensitive ctDNA-based assays and epigenetic biomarkers have demonstrated the ability to identify CRC at asymptomatic or early stages, potentially improving screening uptake and compliance. Furthermore, we explore how liquid biopsy enables dynamic monitoring of treatment response and clonal evolution, facilitating the timely identification of resistance mutations and supporting personalized therapy adjustments. Innovations in multi-omics integration, artificial intelligence, and ultra-sensitive sequencing technologies are also discussed as pivotal advancements that enhance the clinical utility of liquid biopsy. Despite significant progress, the widespread adoption of liquid biopsy faces several hurdles, including assay standardization, sensitivity for low-shedding tumors, regulatory approval, and cost-effectiveness. Continued research, validation in large prospective trials, and harmonization of testing protocols are essential to overcome these challenges. Ultimately, liquid biopsy holds the potential to become a cornerstone of precision oncology in CRC, enabling earlier intervention, more tailored treatment strategies, and improved patient outcomes.

PMID:40905096 | DOI:10.1002/dc.70009

There are no more articles
❌