Towards accurate differential diagnosis with large language models.
利用大型語言模型進行準確的鑑別診斷。
Nature 2025-04-09
這項研究介紹了Articulate Medical Intelligence Explorer (AMIE),一個大型語言模型,旨在提升醫療診斷推理。研究評估AMIE在生成鑑別診斷的有效性,並與傳統方法比較。二十位臨床醫師針對302個複雜案例進行評估,結果顯示AMIE的準確率達59.1%,遠超過未經協助的醫師33.6%。使用AMIE的醫師準確率為51.7%,也高於僅依賴傳統資源的醫師。這些結果顯示AMIE有潛力提升診斷準確性,並強調在真實環境中進一步評估的必要性。
PubMedDOI♡
站上相關主題文章列表
[Making a diagnosis with the aid of the internet in times of large language models].
在大型語言模型時代借助互聯網進行診斷。
Ned Tijdschr Geneeskd 2024-03-26
Can Artificial Intelligence Mitigate Missed Diagnoses by Generating Differential Diagnoses for Neurosurgeons?
人工智慧是否能透過為神經外科醫師生成不同診斷來減輕漏診情況?
World Neurosurg 2024-05-17
Comparative Analysis of Diagnostic Performance: Differential Diagnosis Lists by LLaMA3 Versus LLaMA2 for Case Reports.
LLaMA3 與 LLaMA2 在病例報告中的診斷性能比較分析:鑑別診斷清單。
JMIR Form Res 2024-11-19
Multiple large language models versus experienced physicians in diagnosing challenging cases with gastrointestinal symptoms.
多個大型語言模型與經驗豐富的醫生在診斷具有挑戰性的腸胃症狀病例中的比較。
NPJ Digit Med 2025-02-05
這項研究比較了大型語言模型(LLMs)與人類腸胃科醫生在挑戰性腸胃病例中的診斷能力。研究分析了67個病例,發現LLM Claude 3.5 Sonnet的準確率達76.1%,明顯高於腸胃科醫生的45.5%。這顯示先進的LLMs能成為腸胃科醫生的有力工具,提供高效且具成本效益的診斷支持。
PubMedDOI
Evaluating Large Language Model Performance to Support the Diagnosis and Management of Patients with Primary Immune Disorders.
評估大型語言模型在支持原發性免疫疾病患者診斷和管理中的表現。
J Allergy Clin Immunol 2025-02-16
Human-AI collaboration in large language model-assisted brain MRI differential diagnosis: a usability study.
大型語言模型輔助腦部 MRI 鑑別診斷中的人類與 AI 協作:一項可用性研究。
Eur Radiol 2025-03-07
Towards conversational diagnostic artificial intelligence.
朝向對話式診斷人工智慧。
Nature 2025-04-09
這項研究介紹了AMIE(Articulate Medical Intelligence Explorer),一個專為醫學診斷對話設計的AI系統。AMIE透過自我對弈學習,提升在各種醫療情況下的表現。在一項隨機雙盲研究中,AMIE與初級保健醫生進行評估,結果顯示其診斷準確性超越醫生。不過,研究也指出使用同步文字聊天的限制,這在臨床上不常見。雖然前景看好,但AMIE在實際醫療環境中的應用仍需進一步研究。
PubMedDOI