Achieving Equity in Generative Artificial Intelligence in Cardiovascular Medicine

Artificial intelligence (AI), specifically generative AI, presents opportunities to advance equity through transformative practices in cardiovascular care. In this review, we describe how generative AI models risk perpetuating inequities through representation bias, measurement and labeling bias, deployment bias, and contextual issues. We propose a framework for equitable generative AI that addresses the bias types using design principles of representation and contextualization, transparency, accountability, and community engagement supported by operational strategies of digital equity and continuous recalibration. We explore how generative AI has the potential to improve access to specialty care, enhance patient-centered care, support health literacy, and broaden clinical trial participation. Finally, we encourage and propose governance and oversight of generative AI models across all stages of the AI lifecycle.