In this episode, we speak with Sarah Gebauer MD anesthesiologist and RAND Senior Researcher, who discusses the critical questions facing healthcare professionals as AI becomes integrated into clinical practice. She's the author of "Machine Learning for MDs" newsletter  and published research on physician attitudes toward AI including the BMJ Evidence-Based Medicine article Survey of US physicians' attitudes and knowledge of AI. Her company Validara Health works on evaluation frameworks for healthcare AI implementation.

Physicians have been using AI for EKG interpretation for decades without fully understanding the algorithms, highlighting that transparency should focus on appropriate usage rather than complete technical knowledge. Most current AI tools operate as Software as a Service rather than regulated Medical Devices, while the FDA struggles to keep pace with rapid AI development. Despite their challenging history with EHR implementation, physicians show strong interest in learning about AI when they believe it will help patients.

The medical liability landscape remains uncertain until legal precedents are established through jury awards, making documentation of clinical decision-making crucial when using AI as additional information alongside other clinical data. Traditional machine learning evaluation metrics often fail to predict real-world clinical performance, where workflow integration and clinician experience prove more important than laboratory results. For professional development, busy physicians benefit most from resources that push information directly to them, such as newsletters and targeted social media follows, rather than formal courses requiring active searching.

Some places to follow along with AI in healthcare:

Machine Learning for MDs newsletter

TLDR AI newsletter

a16z healthcare

If you enjoy the show, please leave a ⭐⭐⭐⭐⭐ rating on Apple or a 👍🏻 on YouTube. Subscribe via the Website. Forward to a friend today!