In This Section      
 

Applying Machine Learning to Population Health at JHHC

AI at JHHC

If you use any of today’s social and entertainment services, you likely benefit from machine learning. This technology processes data about you and your content consumption and recommends other content you may enjoy.

Applying machine learning in health care helps Johns Hopkins HealthCare (JHHC) understand and improve population health. Led by Romy Hussain, senior director of Healthcare Economics, JHHC employs a cross-functional department dedicated to finding advanced analytical solutions to promote stronger and more effective patient care. Romy calls her team “a decision intelligence strike force” that brings together applied mathematics, data science, economic theory, and managerial science into a single unit that helps JHHC—and the larger Johns Hopkins Medicine—use data responsibly to make better, more scalable actions.

Machine learning algorithms help find patterns in enormous amounts of data, sometimes uncovering correlations otherwise unknown. Romy, Dr. Xiaonan Zhang, data scientist, and the JHHC Data Science team have created cutting-edge machine learning models to discover deeper insights and predictive indicators. These models have demonstrated high accuracy in predicting hospital readmission risk and length of stay, as well as disease escalation within patients. Through collaboration with clinician partners, JHHC has used these models to develop initiatives that improve the quality of health care. In doing so, doctors and nurses can care for their patients more effectively. One of the machine learning model, Callisto, identifies which health plan members are the best fit to benefit from high-touch care management interventions that may limit future health care utilization. Romy’s team’s work in the area also found that the greatest predictors of adverse health outcomes were social determinants, rather than morbidity.

Romy has also been applying machine learning to the coronavirus (COVID-19) pandemic. Her team’s modeling identified JHHC health plan members most at-risk for COVID-19 and complications and enabled JHHC to quickly contact them with safety reminders and support. Romy recently presented her work around COVID-19 modeling at the Ai4 conference on artificial intelligence and machine learning.

Using this technology, JHHC is gaining insights into population health like never before and applying them to create initiatives that decrease health care spending and utilization, while improving patient outcomes.