Governance Framework for Safe and Ethical Implementation of Artificial Intelligence in Surgery: A Modified Delphi...

Area of Science:

• Surgical Innovation
• Medical Informatics
• Artificial Intelligence in Medicine

Background:

• Artificial intelligence (AI)-enabled clinical decision support systems (CDSS) show expert-level performance in specific medical tasks.
• Significant implementation gaps and ethical, privacy, and legal concerns hinder AI CDSS integration in surgery.
• Need for governance frameworks to ensure responsible AI adoption, prevent care variation, and mitigate algorithmic bias in surgical settings.

Purpose of the Study:

• To establish a systematic, evidence-based, and consensus-driven framework for the ethical, effective, and sustainable adoption of AI-enabled CDSS in surgery.
• To provide foundational guidance for navigating the complexities of AI implementation in surgical practice.

Main Methods:

• Conducted a systematic literature review of major databases (PubMed, Cochrane Library, Medline, Embase) up to 2024.
• Identified key AI governance themes from 80 full-text articles.
• Refined candidate items through a multi-round expert panel consensus process using a modified Delphi approach.

Main Results:

• Thematic analysis revealed four core AI governance themes: Technical Prerequisites and Model Design, Clinical Implementation and Human Factors, Ethics, Safety, and Trustworthiness, and Bias, Fairness, and Equity.
• A 19-item framework was developed through expert panel consensus.
• The framework provides essential guidance for AI governance in surgery.

Conclusions:

• The presented framework offers crucial guidance for the safe and ethical implementation of AI-enabled CDSS in surgery.
• Addressing the four core themes facilitates responsible AI adoption and advances data-driven surgical practice.
• Mitigating potential risks is key to accelerating the transition to advanced surgical care.