Mathematica’s multidisciplinary team joins a short list of seven finalists in the Centers for Medicare & Medicaid Services (CMS) Artificial Intelligence (AI) Health Outcomes Challenge. Mathematica will advance to the final round of the multistage competition, which was launched last year with more than 300 entities proposing AI solutions for predicting patient health outcomes. The competition entries aim to revolutionize health care and are designed for potential use by the CMS Innovation Center. Partners include The Health Collaborative, a leading health information exchange, and the Patient Advocate Foundation.
“We’re excited to advance to the finals of this competition and use our AI expertise to help predict and address unplanned and adverse health events, particularly as we continue to grapple with the COVID-19 pandemic,” said Alex Bohl, director of data science and team leader. “Our marriage of AI sophistication with human-centered design ensures that our predictions are accurate, are accepted by clinicians and patients, and can help improve patient care and clinical outcomes.”
In this last stage of the competition, Mathematica will extend its proprietary algorithms for predicting unplanned hospital and skilled nursing facility admissions and adverse events by introducing more granular patient cohorts and additional outcomes of clinical interest. In addition, Mathematica will deepen its direct engagement with clinicians and patients to ensure that its AI is ethical, is understandable, and drives measurable improvements in patient care. CMS will announce the grand prize winner (who will receive up to $1 million in prize money) and runner-up (who will receive up to $230,000 in prize money) by the end of April 2021.
The CMS AI Challenge attracted innovators from all sectors—not just from health care—to harness AI solutions to predict health outcomes. To select the seven finalists, CMS conducted a rigorous review that included checks for compliance with application requirements and an assessment by AI experts. Submissions aimed to forecast a variety of outcomes, including unplanned admissions related to heart failure, pneumonia, chronic obstructive pulmonary disease, and other high-risk conditions; submissions also predicted adverse events such as hospital-acquired infections, sepsis, and respiratory failure. In Stage 2, finalists will be required to address implicit algorithmic biases that affect health disparities in their submissions.
CMS evaluated each submission based on the model’s performance—specifically, accuracy and the clarity of the visual displays showing how clinicians could use the model forecasts to improve patient care and outcomes. Clinicians from the American Academy of Family Physicians, a CMS partner in the AI Challenge, reviewed and evaluated the visual displays. A panel of CMS senior leaders reviewed the assessments of the expert evaluation panel, and selected the seven finalists, who will each receive $60,000 in prize money and compete for the grand prize and runner-up awards.
The CMS AI Health Outcomes Challenge was launched in March 2019 by the CMS Innovation Center in collaboration with the American Academy of Family Physicians and Arnold Ventures.
More information on the CMS AI Health Outcomes Challenge is available at go.cms.gov/AI.