927 papers
This paper presents an AI-driven approach for the automated 3D segmentation and visualization of 11 anatomical structures in the lower forearm, aiming to enhance musculoskeletal ultrasound proficiency and serve as a bedside teaching tool.
This paper proposes a blockchain-based framework utilizing Non-Fungible Tokens (NFTs), smart contracts, and cryptographic techniques to enable verifiable ownership, controlled access, and forensic auditing of encrypted medical datasets, thereby addressing data sharing barriers in biomedical research.
This paper proposes a unified two-stage deep learning cascade that utilizes an EfficientNetV2-S model with asymmetric loss to detect 28 retinal diseases from the RFMiD dataset and subsequently employs a Platt-calibrated uncertainty-aware gate to route diabetic retinopathy-positive cases to a second stage for ordinal severity grading on the APTOS-2019 dataset, achieving high performance while addressing the significant degradation in recognition accuracy caused by multi-disease co-occurrence.
The paper proposes EBC-FL, an embodied brain-cerebellum federated learning framework for satellite-assisted low-altitude wireless networks that optimizes the trade-off between safety-critical reliability, semantic awareness, and upload volume by integrating ground-based semantic memory, MEO satellite coordination, and lightweight local policies.
PhenoAgent is a privacy-preserving, agentic LLM framework that utilizes structured query decomposition, self-correction, and multi-model consensus to accurately extract clinical phenotypes from free-text electronic health records, significantly outperforming traditional baselines while matching physician-level performance on Japanese hospital datasets.
This study compares human and AI-assisted thematic analysis in surgical qualitative research, finding that while both methods identify similar overarching themes, human-generated codebooks yield higher inter-coder reliability due to clearer thematic boundaries, suggesting a hybrid approach combining AI efficiency with human refinement may be optimal.
This paper proposes the Modality-Missing Robust Representation Network (MMR-Net), a deep learning framework trained on the public IXI dataset that utilizes modality-specific encoders, a shared latent space, and a modality dropout strategy to maintain robust performance in brain MRI age regression and sex classification tasks despite missing imaging sequences.
This study develops and evaluates an interpretable, multi-model NLP framework that effectively detects adverse drug reactions and assesses treatment efficacy from large-scale patient reviews, demonstrating that traditional machine learning approaches outperform transformer-based zero-shot classification in this domain.
This study benchmarks 16 large language models on real-world emergency department triage data, finding that while structured prompting can achieve substantial agreement with human nurses for severity classification, most models exhibit limited accuracy, poor sectoral assignment, systematic overconfidence, and non-deterministic behavior, indicating they are not yet ready for clinical implementation without further validation and improvements.
This study demonstrates that an iteratively optimized, accessible AI tool utilizing meta-prompting and domain-specific guidelines significantly outperforms both baseline AI models and human peer reviews in providing structured, high-quality feedback for plastic surgery manuscripts, thereby offering a scalable solution to support researchers in resource-limited settings.