临床决策支持系统
背景(考古学)
公司治理
决策支持系统
风险分析(工程)
自动化
计算机科学
知识管理
临床治理
过程管理
芯(光纤)
有意义的使用
临床决策
健康信息学
管理科学
人工智能
风险治理
数据科学
患者安全
临床实习
风险评估
梅德林
决策分析
人工智能应用
作者
Huzaifa S Wasanwala,Jacob T. Kannarkat,David Westfall Bates
标识
DOI:10.1093/jamia/ocag155
摘要
OBJECTIVE: Artificial intelligence (AI)-enabled clinical decision support (CDS) challenges core assumptions of the Food and Drug Administration's (FDA) recently updated CDS guidance, including the concepts of independent clinician review of outputs and manufacturer-intended use for exempting certain CDS tools from medical device regulation. MATERIALS AND METHODS: We examine the FDA's January 2026 CDS guidance in the context of the evolving landscape of AI-enabled CDS tools, clinician interaction studies with AI-enabled CDS, and AI governance frameworks. RESULTS: Although the FDA's framework distinguishes non-device CDS from regulated software, AI-enabled CDS may synthesize and transform clinical information in ways that are difficult to verify during routine care, increasing clinician susceptibility to automation bias. DISCUSSION: Incorporating pragmatic triggers into existing governance frameworks can help identify scenarios with a heightened risk for clinical decision influence. CONCLUSION: Oversight calibrated to real-world clinical decision influence is needed to strengthen practical safety standards for AI-enabled CDS.
科研通智能强力驱动
Strongly Powered by AbleSci AI