5 Topic Commentaries

What it takes to integrate clinical AI

• 

  John Halamka, MD

  Biomedical Informatics; Emergency Medicine

  •

  Mayo Clinic

  Handing a data model and an AI model to a physician is not going to get its use and impact. It has to be fully integrated into their workflow.

  Source
• 

  Cui Tao, PhD

  Biomedical Informatics

  •

  Mayo Clinic

  If AI is going to become a part of real clinical care, we need rigorous validation, a governance framework, workflow integration and translational maintenance. And that translation from innovation to responsible implementation is exactly where the field is heading next.

  Source
• 

  Suchi Saria, MSc, PhD

  Biomedical Informatics

  •

  Johns Hopkins University

  Start with: Is it solving a problem? Is it a problem that’s important to solve? How do you evaluate if it is successful? Do you have the resources? Do you understand what risks, if any, it poses? Do you have a mitigation plan that continually monitors and improves performance?

  Source
• 

  Nigam Shah, MBBS, PhD

  Biomedical Informatics

  •

  Stanford Health Care

  AI can augment the practice of physicians and other health care providers, but it’s not helpful unless it’s embedded in their workflow and the information the algorithm is using is in a medical context.

  Source
• 

  Robert M. Califf, MD, MACC

  Cardiology

  •

  U.S. Food and Drug Administration

  Stakeholder education, including healthcare professionals, patients, developer community, and regulators, will be essential to support adoption.

  Source