The Croatia Consensus: Establishing International Best Practices for the Validation and Safe Implementation of Artificial Intelligence in Medically Assisted Reproduction (MAR)
Hickman et al., 2025
As artificial intelligence (AI) becomes increasingly common in fertility clinics, there's an urgent need for guidelines to ensure these tools are safe and effective. This study brought together 148 international experts to create the first comprehensive framework for validating and implementing AI in assisted reproduction. The consensus addresses critical issues including how to test AI systems across multiple clinics, ensure they work fairly for all patient populations, and maintain transparency in AI decision-making. Key recommendations include requiring external validation at multiple centers before clinical use, conducting demographic audits to prevent bias, and ensuring AI systems provide clear explanations of their recommendations. The framework also emphasizes the need for continuous monitoring after implementation to detect any decline in performance over time. This consensus provides fertility clinics, regulatory bodies, and AI developers with a roadmap for responsible AI integration that prioritizes patient safety and clinical outcomes.