CMS Launches $1M AI Development Challenge Aimed at Predicting Unnecessary SNF Admissions

The Centers for Medicare & Medicaid Services (CMS) on Wednesday announced an open challenge to tech providers, dangling $1 million for the algorithms that can best predict potential admissions to skilled nursing and other health facilities.

With the Artificial Intelligence Health Outcomes Challenge, CMS hopes to entice tech firms and individuals to compete against each other to develop artificial intelligence that can comb through Medicare fee-for-service data for potential predictive patterns.

“The Artificial Intelligence Health Outcomes Challenge is an opportunity for innovators to demonstrate how artificial intelligence tools — such as deep learning and neural networks — can be used to predict unplanned hospital and skilled nursing facility admissions and adverse events,” CMS administrator Seema Verma said in a statement announcing the initiative. “For artificial intelligence to be successful in health care, it must not only enhance the predictive ability of illnesses and diseases, but also enable providers to focus more time with patients.”


In addition to predictive analytics, CMS wants its private-sector tech partners to come up with solutions that could help Medicare providers streamline the quality improvement process, while also beefing up the validity of the government’s own internal quality measures.

“The full power of AI can be truly unleashed when providers understand and trust data- and AI-driven predictions,” Verma said in a video on Twitter.

CMS will initially select 20 participants to compete in the first stage of the challenge, with five making it to second stage and receiving an automatic $80,000 reward. From there, the eventual winner will receive $1 million, with the runner-up going home with $250,000.


“Participants will analyze large health care data sets and develop proposals, AI-driven models, and frameworks that accurately predict unplanned hospital and SNF admissions and adverse events,” CMS wrote on its website for the initiative.

Funding for the cash prizes, which will total up to $1.65 million, came from the American Academy of Family Physicians and the Laura and John Arnold Foundation. Interested developers have until June 18 to submit a formal application with the government.

Predicting whether or not a given individual will require more intensive care has long been a goal of software providers in the skilled nursing and broader health care marketplaces. Solutions such as Real Time Medical — which recently raised $10 million to expand its analytics platform — automatically flag changes in skilled nursing resident conditions, serving as an early warning for nurses that a person may need additional attention.

For individual operators, the goal is to prevent costly hospital readmissions, keeping residents in the SNF as long as possible — and given the increasing medical capabilities of the average nursing home, physicians and nurses can often perform the kinds of higher-level procedures once reserved for hospitals within the SNF walls. And with CMS programs such as the value-based purchasing (VBP) initiative placing automatic penalties on facilities that don’t lower hospital readmissions, the pressure to provide more care at SNFs has been intense.

But the CMS AI initiative marks a step toward applying data analytics to the overall Medicare beneficiary population, potentially presaging a future where the government bases reimbursement policy on complex software calculations

“A data revolution is beginning in health care,” Verma said.

Companies featured in this article: