Patient safety governance in artificial intelligence-enabled healthcare: a systematic review and the artificial intelligence patient safety governance model

Abstract

Artificial intelligence (AI) is rapidly becoming part of everyday healthcare practice, helping hospitals and clinics diagnose diseases, recommend treatments, monitor patients and manage medical data. While these technologies offer significant opportunities to improve the quality and efficiency of healthcare, they also introduce new risks that could affect patient safety. This study examines how patient safety can be safeguarded in the era of AI-enabled healthcare. Using a systematic literature review (SLR) guided by the PRISMA (Preferred reporting items for systematic reviews and meta-analyses) approach, the study identifies recurring challenges related to unclear accountability in AI-supported decisions, overreliance on automated recommendations and operational monitoring that may expose patients to potential harm if not properly managed. Building on these insights, the study proposes the AI patient safety governance model (AIPSGM), a structured framework designed to ensure that patient safety is maintained throughout the lifecycle of AI deployment in healthcare institutions. The model highlights five governance pillars: safe AI design and validation, human oversight and clinical accountability, data governance and cybersecurity protection, operational monitoring and risk management and ethical and regulatory governance. In addition, the study introduces the AI Patient safety governance index (AIPSGI) as a practical tool that translates governance principles into measurable indicators that can help healthcare institutions assess their readiness for safe AI adoption. The findings suggest that patient safety in the AI era cannot be achieved through technological accuracy alone but requires continuous governance that integrates technological organizational, ethical and regulatory safeguards. Therefore, this study contributes to the emerging field of responsible AI in healthcare and concludes that patient safety in the AI era depends not only on technological capability but also on the strength of governance structures that guide how these technologies are designed, implemented and monitored.