<![CDATA[Artificial intelligence (AI) is rapidly transforming healthcare delivery in Indonesia. However, the responsible governance of AI systems especially in clinical settings remains underdeveloped. This study aims to identify and prioritize measurable governance indicators for AI in Indonesian healthcare through a Delphi based expert consensus process. A three round modified Delphi method was employed, engaging 30 interdisciplinary experts from healthcare, IT, cybersecurity, ethics, law, and patient advocacy. The process began with 40 indicators drawn from global frameworks (WHO, EU AI Act, ISO/IEC 42001, NIST RMF) and national references (UU PDP, SATUSEHAT). Experts rated each indicator on a 1–9 Likert scale across two iterative rounds. Consensus was defined as median ≥7 and IQR ≤1.5 using RAND/UCLA criteria.Out of 40 indicators, 24 achieved consensus. High priority indicators included clinical safety metrics (e.g., AUROC), data privacy compliance (PDP Law documentation), system integration (SATUSEHAT compatibility), and cybersecurity readiness (incident response plans). Transparency related indicators (e.g., training data summaries, model cards) failed to reach consensus, suggesting institutional gaps in AI explainability. The Delphi process underscored the importance of participatory governance, stakeholder trust, and contextual adaptation of international standards. Consensus indicators reflect domains where operational familiarity and regulatory anchors already exist, while non consensus areas highlight the need for capacity building and clearer guidelines. This study delivers a validated, measurable governance framework to guide responsible AI adoption in Indonesian healthcare. It supports policymaking, institutional audits, and procurement strategies aligned with both local regulation and global standards. Future work should pilot these indicators and expand their use in health system assessments and continuous governance improvement.]]>
Copyrights © 2025
