Chicago ACO Uses AI to Reach Most Vulnerable COVID-19 Members

The ACO network, which includes Rush University Medical Center and Sinai Health System, uses machine learning solution to identify and risk stratify members with social and medical risks.

In the time of COVID-19, how can healthcare systems identify their most vulnerable patients and communicate with them to mitigate risk and unnecessary use of the emergency department? Medical Home Network has turned to artificial intelligence (AI).

By cross-referencing 122,000 patients with specific social and medical issues, the technology identifies those at highest risk for severe complications so care managers can proactively reach out to them. This approach could provide a model for other health system to follow to pinpoint their own vulnerable populations.

"When COVID-19 came, one of our concerns was doing outreach and education, particularly for our most vulnerable members," the network's Chief Medical Officer Art Jones, MD.

The Chicago-based organization is affiliated with the Medical Home Network Accountable Care Organization (ACO) that includes Rush University Medical Center, Sinai Health System, La Rabida Children’s Hospital, ten Federally Qualified Health Centers, and their physician practices. It is working with Austin, Texas-based ClosedLoop, which provides the AI-based predictive model.

One advantage Medical Home Network had is that long before COVID-19 became a public health issue, members had already completed an assessment related to social determinants of health, explains Jones. Among the social risk factors identified: Homelessness, living alone, and those who did not have a family member or friend to help if they became ill. The AI system is cross-referencing this data against medical vulnerabilities, such as age, diabetes, and lung disease. 

Jones says the data analysis was expected to be completed yesterday, and outreach by care managers within each participating organization would begin today. Because of social distancing restrictions, all outreach will be conducted by phone.

Care managers will educate patients about how they can lower their risk of catching the infection, early symptoms, and what to do if they become sick, says Jones. One key element of the initiative, he says, is to let patients know they can reach out to their care manager or primary care provider versus visiting the emergency room.

"One fear with the pandemic is that is that people will just rush to the emergency room when they develop relatively minor symptoms, instead of contacting their care managers and their primary care practices," says Jones.

"In addition to those initial outreach calls," he says, "we'll try and do periodic wellness checks on those vulnerable patients."

Other healthcare systems can replicate this process by using a "machine learning type of approach to their medical claims data" to identify priority patients, says Jones. "Also, some medical providers are beginning to collect [information about] social determinants of health … in their EHR (electronic health record). So to the extent that the provider has then collected information, they could duplicate this approach. If not, their health plan could use claims data to identify people strictly from a medical perspective that are [at risk]."

Another advantage Medical Home Network has, says Jones, is that because patients are already familiar with their care managers, they are more likely to answer the phone.

"The plan has delegated care management down to our practices," says Jones. The advantage is that you have the care managers as part of the care team who build face-to-face trusting relationships with their patients. [Patients are] much more likely to pick up the phone when they see in their caller ID that the call coming from their doctor's office than if it's coming from a strange number, like a health plan number."