Why Ascension’s many EHRs became an AI advantage

Advertisement

Running multiple EHRs forced St. Louis-based Ascension to build one clean, unified record across all of them. Raj Mohan, the system’s chief marketing, conversational and digital experience officer, said that work is now paying off with AI.

It pushed the system, years ago, to pull data from those EHRs into the cloud and harmonize it so the same patient is recognized across EHR instances. That clean, unified data, available on a single platform through APIs, now underpins Ascension’s website, app and AI tools.

Mr. Mohan credited Ascension’s own data staff with harmonizing it into a unified record.

“It’s really the teams at Ascension that do the hard lift itself,” he said. “The platform is just the foundation. The experience is ours to create.”

Those experiences include referral follow-up. When a clinician places an order or referral in the EHR, Ascension pulls it into its CRM system, reaches out to the patient within a set period and checks a couple of days later whether an appointment has been made.

The contact center is where this work is most visible, and it’s Mr. Mohan’s newest responsibility. He recently added “conversational” to his title when the contact center joined his portfolio, calling it “a nod to where we want the experience to go versus the channel it happens on.”

There, Ascension’s AI scheduling assistant, Grace, books new primary care appointments and verifies, cancels and changes existing ones. About three months in, it handles roughly 20% of more complex scheduling calls, such as those involving new appointments, and as much as 65% of simpler ones. AI-enabled conversations already number in the tens of thousands each week, though Mr. Mohan said the system is still in the early stages. Beyond satisfaction scores that are better than interactive voice response and comparable to human agents, Ascension tracks whether callers completed what they called to do.

Health systems running a single EHR have an easier starting point, he acknowledged. Still, organizations that don’t combine clinical data with other information about consumers are left with medical records alone, “which is only a fraction of who a person is,” he said.

That’s why Ascension says “consumer” rather than “patient.” Mr. Mohan traced the choice to a research session in which one person told the health system, “Don’t call me a patient,” explaining that they were only a patient for the seven to 12 minutes spent in front of the physician and wanted to be treated with the respect of a consumer the rest of the time.

Ascension is now working on how to better understand patients’ needs, wants and behaviors beyond what’s in the health record or social determinants of health data. That matters more as the system designs AI and conversational experiences, Mr. Mohan said, because much of what people pick up on in human interactions are “cues that are not written down anywhere.”

Advertisement

Next Up in Digital Health

Advertisement