Artificial intelligence (AI) software can improve patient care, but a key market failure is preventing its effective development and diffusion.
Well-developed and validated AI software systems built outside of electronic health record (EHR) systems that could truly advance patient care have had limited adoption. Only a small number of health systems have developed successful and integrated AI software, and broad adoption of these validated tools is essentially nonexistent. In comparison, poorly developed and minimally validated AI software built and deployed within EHRs are widely adopted despite serious quality concerns.
Breaking down the unit costs of AI software exposes why current incentives lead to this market failure. We offer a reimbursement framework and policy intervention that rectifies this problem by better aligning AI software adoption with emerging best practices.
Create a display name to comment
This name will appear with your comment
