• CMS Names Winners of AI Health Outcomes Challenge

    AAFP Played Essential Role in Judging Entries, Supporting Competition

    May 24, 2021, 3:38 p.m. Cindy Borgmeyer ― Harnessing the power of artificial intelligence to help predict patient health outcomes and, hopefully, avoid a variety of adverse health-related events took a big leap April 30, when CMS announced that data science platform ClosedLoop.ai had been selected the winner of its AI Health Outcomes Challenge.

    As the sole medical professional organization sponsoring the competition, the AAFP ― in partnership with CMS’ Center for Medicare and Medicaid Innovation, also known as the CMS Innovation Center, and the Laura and John Arnold Foundation ― played a key role in reaching this milestone after a journey that’s been years in the making.

    AAFP Continues Investment in Medical Informatics

    In October 2018, the AAFP Board of Directors approved a multiyear special informatics project focused on shaping the future of health IT for family physicians by leveraging new technologies.

    Under the direction of AAFP Vice President and Chief Medical Informatics Officer Steven Waldren, M.D., M.S., the Academy has made major progress toward that goal. Along with collaborating to launch the AI Challenge in March 2019, it established the AAFP Innovation Lab to partner with innovators in evaluating tech solutions for possible deployment in family medicine practices, work that continues today.

    Waldren said he suspects it was that very commitment to innovation that was a key factor in CMS asking the AAFP to participate in the challenge ― but it likely wasn’t the only reason.

    “Just my conjecture here, but I think it was due to a few things,” he told AAFP News. “First, our work and advocacy with CMMI around alternative payment and our being seen as open to and pushing innovation. Second, I think it is due to our long history of work in health IT.

    Story Highlights

    “I also think it was because family medicine has a large scope and the greatest window to intervene when an AI solution would identify an opportunity.”

    AI Challenge Drew Extensive Participation

    The multistage competition engaged innovators from all sectors ― not just health care ― to harness AI solutions capable of predicting health outcomes for possible use in CMMI’s payment and service delivery models.

    “Clinicians are eager to use the latest innovations to better help identify patients at risk, provide higher quality care, and improve health outcomes,” said CMS Acting Administrator Liz Richter in the April 30 press release. “The use of artificial intelligence has the potential to achieve these aims by providing important information to clinicians that may be helpful in providing higher quality care.”

    Styled as an opportunity for innovators to demonstrate how AI tools can be used to predict unplanned hospital and skilled nursing facility admissions and adverse events ― a $200 billion problem that impacts nearly 32% of Medicare beneficiaries ― the challenge prioritized explainable solutions to help front-line clinicians comprehend, appreciate and trust AI-driven data feedback to improve quality of care.

    Based on established criteria, a group of data science experts, health care data specialists, and practicing clinicians evaluated the more than 300 initial applications, winnowing them down to 25 entries that advanced to Stage 1 of the challenge in October 2019. In this stage, participants designed and tested their proposed solutions using specified Medicare claims data sets. Seven of those 25 participants were chosen to move on to Stage 2 in October 2020, during which they further refined their algorithms and solutions using additional CMS data sets.

    According to Waldren, the AAFP was involved in virtually every stage of the process. “We participated in calls with CMMI and their contractor MITRE during the project,” he explained. “We also identified family physicians to serve as judges during both stages of evaluations.

    “Specifically, our members judged the clinical merits and usability of the applications. Notably, 40% of the scoring was dedicated to how the AI solution interfaced with clinicians.”

    In addition to naming Austin, Texas-based ClosedLoop.ai the winner of the contest, the agency announced that nonprofit Geisinger Health, based in Danville, Pa., had been selected runner-up. ClosedLoop.ai will be awarded $1 million in prize money, while Geisinger will receive $230,000. In addition, each of the seven finalists was awarded $60,000.

    In the CMS release, Academy EVP and CEO Shawn Martin pointed to the overall progress made to date in exploring the use of smart technology to aid physicians and benefit patients.

    “We are excited about the early successes and great potential of artificial intelligence to dramatically improve health outcomes, reduce administrative burden, and create smarter health IT,” he said. “We were pleased to partner with the CMS Innovation Center on the AI Health Outcomes Challenge and to bring family physicians into the evaluation of the contestants to help identify the winners.

    “We look forward to seeing the winning, and all of the great, solutions in the market.”