Tag: Medical AI

A new study demonstrates that foundation model-based multiple instance learning can predict clinical outcomes in head and neck squamous cell ...

A new deep learning algorithm called PaliConvLSTM classifies arrhythmias into three risk levels, achieving 99.68% accuracy on standard ECG datasets ...

An agentic framework enables open-source small language models to accurately convert unstructured chest X-ray narratives into standardized anatomical sections for ...

A 588-gram abdomen-mounted robot controlled by a sonographer-trained AI agent acquired clinically interpretable liver ultrasound images in healthy volunteers and ...

A four-round international eDelphi study with 53 experts from 21 countries reached consensus on 132 of 170 items defining bioethical ...

A qualitative study of 45 stakeholders in Shenzhen finds concerns about safety, regulatory clarity, responsibility allocation and cross-sector cooperation in ...

A qualitative interview study of hospital staff identified six ethical themes in AI-assisted clinical practice and led researchers to propose ...

Researchers evaluated domain adaptation and fine-tuning of five large language models for radiology impression generation across 3.1 million reports, finding ...

A new narrative review finds that medical vision–language AI models show strong benchmark results but suffer from inconsistent evaluation reporting, ...

A qualitative study of 26 tuberculosis physicians in Hubei Province, China, identified five priorities for explainable AI, including exclusionary and ...

A diabetes question-answering prototype combines a structured medical knowledge graph, guideline retrieval and a large language model while exposing the ...

A repeated Japanese survey of physicians and the public found that intention to use relatively autonomous medical AI did not ...

A critical narrative review finds that AI improves detection in mammography and colonoscopy but does not improve complex diagnostic reasoning, ...

Six large language models rarely discouraged medically indicated care but became less consistent when recommending unnecessary tests in areas where ...

A hybrid AI framework combining Mamba2 context modeling, frequency-domain refinement and local texture analysis outperformed leading baselines for detecting pulmonary ...

Harvard-led researchers adapted an adult-trained AI model for newborns by training it sequentially on chest x-rays from 30 neonatal intensive ...

A new iScience study introduces CSCC-PAG, a deep learning framework that combines domain-specific pre-training with self-supervised domain adaptation to improve ...

In-depth interviews with 17 neonatologists at a Dutch academic hospital reveal that physicians distinguish trust from confidence and reliance, and ...

Researchers have built a temporal-aware retrieval-augmented generation framework that helps large language models generate safer, evidence-based treatment plans for idiopathic ...

A new vision-language model fuses MRI, diffusion imaging and clinical scores to classify Alzheimer's disease, works even when data is ...