西南医科大学学报2026，Vol.49Issue(4)：409-414,6.DOI:10.3969/j.issn.2096-3351.2026.04.001
生成式人工智能辅助心血管疾病诊疗决策的信任构建与挑战
Trust Building and Challenges of Generative Artificial Intelligence-Assisted Cardiovascular Diagnosis and Treatment Decision-Making
摘要
Abstract
In cardiovascular diagnosis and treatment,generative artificial intelligence(Generative AI)demonstrates significant potential in efficiently generating medical records,providing personalized diagnostic and therapeutic recommendations,and optimizing physician-patient communication.However,its inherent"black-box"nature,unpredictable hallucination phenomena,and complex medical ethics have precipitated a crisis of trust among clinicians and healthcare administrators.This paper systematically explores the specific requirements for AI accuracy,interpretability,and robustness in the cardiovascular field from both technical and non-technical perspectives.On the technical front,the paper analyzes interpretability methods such as attention visualization,thought-chain prompts,and Retrieval-Augmented Generation(RAG),as well as the role of uncertainty quantification in risk warning.On the non-technical front,it emphasizes the critical importance of rigorous clinical validation,cross-jurisdictional regulatory frameworks,and the delineation of liability,as well as human-AI collaboration models and the cultivation of AI literacy among physicians for build-ing trust.The article further summarizes the current technical bottlenecks,data challenges,and barriers to acceptance,and proposes future directions involving technological convergence,collaborative decision-making,and dynamic trust frameworks.Trust is a multidi-mensional systems engineering endeavor that requires the coordinated advancement of technology,institutional frameworks,and humanistic considerations to achieve the reliable implementation of generative AI in cardiovascular clinical practice.关键词
生成式人工智能/心血管疾病/临床决策支持/信任构建/可解释性/人机协作Key words
Generative artificial intelligence/Cardiovascular disease/Clinical decision support/Trust building/Interpretabil-ity/Human-machine collaboration分类
医药卫生引用本文复制引用
张春祥,苏渝杰..生成式人工智能辅助心血管疾病诊疗决策的信任构建与挑战[J].西南医科大学学报,2026,49(4):409-414,6.基金项目
国家自然科学基金(U23A20398) （U23A20398）
国家科技重大专项项目(2024ZD0537707) （2024ZD0537707）
四川省科技计划项目(2025YFRG0005) （2025YFRG0005）
泸州市人民政府-西南医科大学科技战略合作项目(2025LZXNYDPD01) （2025LZXNYDPD01）
西南医科大学研究创业基金(00040155) （00040155）
