Community Health Network prescribes AI to combat COVID-19

AI is about to play a much larger role in identifying individuals who are at risk for contracting COVID-19 in Indiana. Community Health Network (CHN), an accountable care organization (ACO), revealed today that it is beginning to employ an AI platform from Jvion to analyze members’ electronic health care records that are stored in a platform from Epic Systems.

Residing on the Microsoft Azure Cloud, the AI Core platform from Jvion applies machine learning algorithms to structured and unstructured data provided by health care organizations alongside data gathered from publicly available sources. It then makes use of an Eigen Sphere-based artificial intelligence engine used widely to track, for example, consumer behavior, in order to enable machine learning algorithms to map ACO members to specific risk profiles.

Jvion earlier this week added a Vaccination Prioritization Index (VPI) to an existing COVID Community Vulnerability Map it created that maps communities and individuals by their vulnerability to COVID-19 and level of priority for vaccination based on guidelines created by the Center for Disease Control (CDC) and other socioeconomic factors.

CHN is using the AI Core platform as part of a larger social equity initiative the health care organization has launched to reach out to individuals who are part of a demographic that for a variety of historical and economic reasons is often disinclined to consult a physician before having to be hospitalized, CHN executive VP Dr. Patrick McGill, MD said.

The ACO currently has more than 200 locations in Indiana through which it provides medical services. COVID-19 has tended to have a greater impact on certain individuals based on a range of socio-economic factors. Jvion takes those factors into account to identify the patient records of ACO members that health care professionals should reach out to as part of an effort to educate them about their higher level of COVID-19 risk, McGill said. “It’s using ML to assess risk scores,” McGill added.

Similar to how financial organizations assess credit risk scores, McGill said the goal is to eventually expand usage of the Jvion platform to enable the ACO to also identify patients at a higher level of risk for contracting other diseases, such as diabetes. As a value-based health care provider, an ACO typically incurs higher costs when patients who defer care wind up back in the hospital.

Funding for these initiatives is being provided by a foundation the ACO created to explore the impact social determinants have on health care. This initiative with Jvion will hopefully yield actionable data that confirms what in many instances are still theoretical assumptions, McGill noted. Longer-term, that capability could be crucial because many individuals are, for example, putting off cancer screening tests until the COVID-19 pandemic subsides. That might result in a large number of cancer cases being discovered later than they normally would and at roughly the same time, McGill noted.

The AI experiment should also increase the confidence medical personnel can have in employing AI to proactively launch health care initiatives that result in fewer hospitalizations, McGill added.

While the primary mission remains providing better health care sooner, the impact this initiative could have on hospital costs is substantial. As COVID-19 vaccines become more widely available and the holiday season moves further behind us, the number of COVID-19 hospitalizations has started to subside. However, more contagious variants of the virus are starting to spread, which could result in much higher rates of hospitalization before enough members of the ACO are inoculated, McGill said.

Regardless of the immediate impact on hospitalization, McGill said this initiative will serve as a test case for applying AI more broadly to preventative health care to not only reduce costs but also save lives.