Diagnostic Performance of ChatGPT-4o and DeepSeek-3 Differential Diagnosis of Complex Oral Lesions: A Multimodal Imaging and Case Difficulty Analysis.
ChatGPT-4o 與 DeepSeek-3 在複雜口腔病變鑑別診斷的診斷表現:多模態影像與病例難度分析
Oral Dis 2025-07-01
研究發現,口腔醫學專家在診斷帶圖片的臨床案例時,表現比 ChatGPT-4o 和 DeepSeek-3 這兩款 AI 更好。雖然 DeepSeek-3 在純文字診斷上比 ChatGPT-4o 準確,尤其在複雜或發炎案例,但整體來說,專家還是最可靠。加入影像資訊能提升診斷準確度,但 AI 目前還無法完全取代專業醫師。
相關文章PubMedDOI推理
Natural language processing in veterinary pathology: A review.
獸醫病理學中的自然語言處理:綜述
Vet Pathol 2025-07-01
Generative-AI-Based Approaches for Information Extraction from Clinical Notes: A Scoping Review.
基於生成式 AI 的臨床紀錄資訊擷取方法:範疇性回顧
Stud Health Technol Inform 2025-07-01
Leveraging Large Language Models for the Extraction of General and Domain-Specific Signs and Symptoms.
運用大型語言模型萃取一般性與特定領域的徵象與症狀
Stud Health Technol Inform 2025-07-01
Leveraging LLMs for Early Diagnosis: Comparing ClinicalBERT and GPT-4 to Address Diagnostic Uncertainty in the Emergency Department.
運用大型語言模型於早期診斷:比較 ClinicalBERT 與 GPT-4 於急診部門解決診斷不確定性的應用
Stud Health Technol Inform 2025-07-01
Relevance of Grounding AI for Health Care.
人工智慧在醫療保健中「Grounding」的重要性
Stud Health Technol Inform 2025-07-01
大型語言模型像 GPT-4 雖然在醫療領域應用越來越多,但因為缺乏真實世界的感官和動作基礎,容易出現推理錯誤或倫理問題。論文提醒,若把這些 AI 當成知識代理人而非輔助工具,可能會對臨床實務帶來風險。
相關文章PubMedDOI推理
Using Large Language Models for Data Cleaning: An Evaluation of ChatGPT-4o's Performance.
運用大型語言模型進行資料清理:ChatGPT-4o 效能評估
Stud Health Technol Inform 2025-07-01