ChristianaCare

Predictive Analytics and Care Management Reduces COVID-19 Hospitalization Rates Avoiding Nearly $2M in Costs

For people 65 years of age and older, COVID-19 hospitalization rates in the U.S. have been as high as 1,245.7 per 100,000 population, straining the resources and capacity of health systems. ChristianaCare needed to effectively deploy its care management resources, identifying patients with COVID-19 who were most at risk of severe illness and hospitalization. By leveraging its analytics platform and the predictive power of the Healthcare.AI™ solution to provide COVID-19 risk prediction, the organization was able to provide targeted interventions to those most likely to benefit and help patients avoid unnecessary hospitalization.

AI Can Advance Health Equity

Health technology and augmented intelligence (AI) can significantly improve or worsen health equity. Recently, there has been a growing concern that AI is increasing disparity.1 ChristianaCare set a goal to reduce avoidable health disparities. The organization faced many challenges, including inconsistent collection, storage, and use of personal characteristics such as race, ethnicity, and language. Using its data platform and Healthcare.AI™, ChristianaCare now has a single “source of truth” for personal characteristics data. By treating health equity as a goal with the same commitment and focus as it would for other clinical, operational, or financial improvement efforts, the organization is purposefully using AI to achieve health equity.

Machine Learning and Feature Selection for Population Health

Christiana Care Health System (CCHS) had used a machine learning model to inform population segmentation. The initial model used “black box” algorithms to predict risk that care managers didn’t have input on or understand. CCHS leaders and experts wanted an efficient model that they understood and trusted to predict 90-day inpatient admission. CCHS used a feature selection process to build the simplest model possible—and AI insight tools for selecting the best model, setting triggers for action, and explaining how the model worked.