臨床推論AI、完全性はまだ遠い：21のLLMの実力を検証

MedAI Digest (JA)
MedAI Digest (JA)
臨床推論AI、完全性はまだ遠い：21のLLMの実力を検証
Loading
/

新たなベンチマークテストにより、最先端のLLMでさえ臨床的意思決定には十分な能力を備えていないこと、特に鑑別診断において著しい弱点があることが明らかになりました。

Original paper: Large Language Model Performance and Clinical Reasoning Tasks. — JAMA network open. 10.1001/jamanetworkopen.2026.4003

📄 記事を読む

Trending now

Quantitative pathology and APOE genotype reveal dementia risk and progression in Lewy body disease.
APOE Genotype and Protein Pathology Shape Dementia Risk in Lewy Body Disease
Transfer-learning guided design of high-performance conjugated polymers for low-voltage electrochemical transistors.
Machine Learning Accelerates Discovery of Ultra-Low-Voltage Polymers
Soft multiclass feature augmented deep learning to predict tumor origins using cytology or histology whole slide images.
AI Identifies Tumor Origins from Tissue Images with 98% Accuracy
AI-guided multi-omics analysis identifies NPC1-modulated susceptibility to SARS-CoV-2 infection under PM2.5 exposure.
Air Pollution and COVID-19: How NPC1 Mediates SARS-CoV-2 Infection Risk