The CMS AI Health Outcomes Challenge
Submission Deadline is June 18, 2019
Overview - On March 27, 2019, CMS announced a three-stage competition to accelerate artificial intelligence (AI) solutions. This competition is an opportunity for innovators to demonstrate how AI tools, such as deep learning and neural networks, can be used to predict unplanned hospital and skilled nursing facility (SNF) admissions and adverse events. The goal is to provide physicians with actionable data and more accurate predictive capabilities so that they can provide appropriate resources to the highest risk patients, at the right time. CMS is partnering with the American Academy of Family Physicians (AAFP) and the Laura and John Arnold Foundation to award up to $1.65 million to selected participants during the three stages of the Challenge.
The Challenge anticipates that, using CMS claims data for Medicare Part A (hospital) and Medicare Part B (professional services), participants will leverage deep learning AI models to identify unplanned hospital and SNF (skilled nursing facility) admissions and adverse events before they occur. CMS provides a number of examples of the types of outputs that may be focus of AI solutions, including, but not limited to:
- Beneficiary Risk Classification: Under this approach, an AI model would classify risk stratification groups (low, medium, high), and bucket each beneficiary into a risk group.
- Beneficiary Risk Clustering: Under this approach, the AI model would cluster beneficiaries into risk stratification groups.
- Event Prediction Probability Score: Under this approach, the AI model would predict the probability of an event occurring such as an unplanned hospital admission or adverse event and deliver an associated confidence interval justifying the strength or weakness of the prediction.
The Challenge Structure - The Challenge consists of three stages.
- During the Launch Stage, participants will submit an application via ai.cms.gov and provide information about their proposed solution. Up to 20 participants will be selected to advance to Stage 1.
- During Stage 1, participants will design and test their proposed solution using certain Medicare claims data sets. Up to five participants will be selected to advance to Stage 2 and will each be awarded up to $80,000.
- During Stage 2, finalists will be able to request additional Medicare claims data and refine their solutions. The grand prize winner in Stage 2 will be awarded up to $1 million and the runner up will be awarded up to $250,000.
Winners of the Challenge will be chosen based on the following criteria:
- Impact of the Proposed Solution (30%) (including the extent to which the proposed approach is operationally feasible for CMS; the extent to which it is likely to succeed in predicting unplanned hospital and SNF admissions and adverse events; and the whether the participant identified potential roadblocks to implementation.
- Innovation of Proposed Solution (30%)(including the degree to which the proposed design is innovative, creative, and original; the extent to which the participant can demonstrate that the proposed solution can outperform existing approaches; and the exten to which the participant identified other data sets and/or types of information that would be useful to further refine their solutions following the competition?
- AI/Human Collaboration (40%) (including the extent to which the participant explains how the proposed AI tool will work with humans (clinicians and patients) to achieve the desired results; the extent to which the participant identifies strategies and tools to explain AI predictions to clinicians and patients; and the extent to which the participant demonstrates a link between the proposed solution and benefit to the Medicare population and potential impact on current health care practice and delivery methods?
Challenge Timing - The Challenge will run for approximately one year. The Launch Stage will run from March 2019 through June 2019. Stage 1 will run from summer 2019 – fall 2019. Stage 2 will run from winter 2019 through spring 2020 and the winner will be announced in April 2020. (Dates are subject to change.).
The Submission Deadline is June 18, 2019.