HLTH Executive Series: Dr. Brian Anderson of CHAI: Ensuring Safe and Effective AI in Healthcare
Independent quality assurance labs are critical for evaluating health AI models, just like in other sectors.
In this episode, Dr. Brian Anderson, president and CEO of the Coalition for Health AI (CHAI), discusses the organization's initiatives to create a national network of certified labs and a model card (AI nutrition label) to assess AI safety and effectiveness. He highlights the challenges in defining and measuring bias, especially in generative AI, and the importance of transparency and collaboration between industry, government, and academia to build trust. Dr. Anderson also covers the need for provider upskilling in AI literacy and the potential of AI tools, like ambient scribes, to mitigate clinician burnout. Finally, he explains how public access to evaluation reports is key to public trust.
Tune in and learn about the future of AI regulation in healthcare and learn why physicians must educate themselves on AI!
About Dr. Brian Anderson:
Dr. Brian Anderson is a leading voice in health AI as CEO and Co-Founder of the Coalition for Health AI (CHAI), guiding the development of national standards for safe and effective AI in healthcare. Formerly Chief Digital Health Physician at MITRE, he spearheaded crucial research initiatives, including advancements in clinical trials and oncology. An internationally recognized expert, Dr. Anderson speaks frequently on digital health, AI assurance, and interoperability. A Harvard Medical School graduate with an MD (honors) and a BA (cum laude), Dr. Anderson trained at Mass General, practiced at Greater Lawrence Family Health Center, and lives in Boston with his family.Things You’ll Learn:
A significant gap exists in the lack of independent labs to evaluate health AI, as these are already standard practice in other sectors with regulated technologies. The proposed national network of certified labs will fill this gap by providing independent assessments of AI models, promoting trust in their use.
AI model cards are crucial for transparency because they detail the training methodologies and ingredients of AI models. This information helps users, such as physicians, make informed decisions about the tool’s appropriateness for their patients.
Clinicians need upskilling to critically evaluate AI tools and make informed decisions about their use in patient care.
Generative AI applications like ambient scribes have the potential to greatly mitigate physician burnout by streamlining administrative tasks. This can give them more time to focus on their patients and improve their work-life balance.
The creation of quality assurance labs will be a critical first step in AI regulation, helping to bridge the gap between rapidly evolving technology and established safety standards.
Resources:
