Geisinger Chosen as Finalist in CMS Artificial Intelligence Challenge
Geisinger and Medial EarlySign have advanced to the final round of the CMS competition for developing an artificial intelligence solution that can detect adverse events.
Source: Thinkstock
By Jessica Kent
- CMS has selected Geisinger as one of seven finalists in its Artificial Intelligence Health Outcomes Challenge.
In collaboration with Medial EarlySign, the health system used artificial intelligence and machine learning to predict unplanned hospital admissions, readmissions occurring soon after hospital discharge, and healthcare-associated complications. The organizations partnered to develop models that predict the risk of these outcomes using Medicare administrative claims data.
“This partnership enabled cross-disciplinary collaboration where both Geisinger and Medial EarlySign leveraged their strong healthcare and data science expertise to solve problems that can help fundamentally transform the healthcare delivery system,” said Karen Murphy, PhD, RN, Geisinger’s chief innovation officer and founding director of Geisinger’s Steele Institute for Health Innovation.
“The types of predictive models and computer user interfaces developed through the CMS AI Health Outcomes Challenge have enormous potential to improve patient outcomes, enhance clinician satisfaction, and reduce healthcare costs.”
The CMS AI Health Outcomes Challenge launched in 2019, with more than 300 organizations proposing AI solutions for predicting patient health outcomes for potential use by the CMS Center for Medicare and Medicaid Innovation.
Submissions aimed to predict a range of outcomes, including unplanned admissions related to heart failure, pneumonia, chronic obstructive pulmonary disease, and other various high-risk conditions. Submissions also focused on forecasting adverse events like hospital-acquired infections, sepsis, and respiratory failure.
“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,” said CMS Administrator Seema Verma at the time the challenge was announced.
“For artificial intelligence to be successful in healthcare, it must not only enhance the predictive ability of illnesses and diseases, but also enable providers to focus more time with patients. The power of artificial intelligence will truly be unleashed when providers understand and trust the data and predictions.”
In November 2019, Geisinger and Medial EarlySign advanced to Stage 1 of the competition among 25 other organizations. The team noted that their algorithm could effectively reduce the number of unplanned admissions to hospitals and skilled nursing facilities, as well as a range of adverse events.
“Approximately 4.3 million hospital readmissions occur each year in the U.S., costing more than $60 billion, with preventable adverse patient events creating additional clinical and financial burdens for both patients and healthcare systems,” said David Vawdrey, PhD, Chief Data Informatics Officer at Geisinger.
“Together with our partner EarlySign, we have forged a dynamic team that is rapidly developing novel solutions to achieve the Quadruple Aim of improving the patient experience of care, improving the health of populations, reducing cost, and improving clinical care provider satisfaction.”
To determine the seven finalists, CMS evaluated each submission based on the model’s performance and how well the developers visually demonstrated how clinicians could use their models to improve patient care and outcomes.
Clinicians from the American Academy of Family Physicians reviewed and evaluated the visual displays, and a panel of CMS leadership reviewed the assessments and selected the seven finalists.
In the final stage of the competition, finalists will further refine their predictive models while also addressing implicit algorithmic biases that impact health disparities. CMS will announce the grand prize winner and runner-up by the end of April 2021.
“Along with Geisinger, being chosen as a finalist is an accomplishment we are humbled to receive,” said Ori Geva, co-founder, and chief executive officer of Medial EarlySign.
“Geisinger’s deep understanding and commitment to patient care augmented with our machine learning modeling allowed us to excel as a team. This challenge has proven to us the depth and value of our medical AI modeling framework and the power it gives our healthcare clients to handle complex clinical predictive modeling at scale.”
