CMS launches $1.65 million AI challenge

The CMS on Wednesday launched a $1.65 million contest to develop an understandable artificial intelligence tool that can predict patients' healthcare outcomes and adverse events.

The right AI tool could improve quality and lower administrative burdens for doctors, The agency and the American Academy of Family Physicians, which is supporting the challenge, say the right AI tool could improve care quality and lower doctors' administrative burdens.

"One of the most important aspects of this challenge, in addition to driving AI and predictive analytics, is to make sure AI is understandable to physicians," said Adam Boehler, director of the Center for Medicare and Medicaid Innovation during an event Wednesday at Johns Hopkins University's Applied Physics Laboratory in Laurel, Md.

Through AI, a doctor wouldn't have to "wade through the large volume of data" to make an appropriate diagnosis, said Dr. Michael Munger, a family physician in Kansas who is also board chair of the AAFP, at the Laurel event.

However, the challenge doesn't address the cost for a hospital or doctor's office to implement an AI tool, which experts say is a major barrier to implementation.

Another concern is how to harness the vast volumes of clinical data that will be needed to fuel an AI algorithm. Data will need to be gathered securely from patients' electronic health record, and AI insights will also need to be inserted back into an EHR workflow.

Regulations of an AI product are another key sticking point, and one that HHS is aware of.

"To unlock the next generation of medical advances, we have to look at how to monitor the safety and effectiveness of software in or as devices that don't rely on a static algorithm, but instead leverage artificial intelligence to constantly adapt and learn," HHS Deputy Secretary Eric Hargan said during the Academy Health's Datapalooza event Wednesday in Washington.

The FDA in January put together a new program aimed at reviewing digital health devices such as AI technology.

Applications for the $1.65 million Artificial Intelligence Health Outcomes Challenge are due on June 18.