WellSpan Health uses AI for mass data migration among multiple EHRs.
To harmonize information among disparate EHRs and establish confidence in post-acquisition system technology, the health system aligned human process improvement and advanced technology combining AI assistance from DrFirst to streamline data migration, manual input and user adoption.
Artificial intelligence played a significant role in helping WellSpan Health facilitate a transition between various electronic health records systems that its facilities were using, bringing efficiency and improving experiences for patients and clinicians.
In a presentation for the HDM KLASroom, WellSpan Health’s Chief Medical Information Officer, Robert M. Lackey, MD, and Donald “Chip” Gerhart, the system’s manager of pharmacy clinical informatics, discussed how AI helped deliver a smooth EHR transition for clinicians.
After a large merger, WellSpan Health had many facilities, five EHRs and a patient-satisfaction problem, Lackey and Gerhart said. Patients didn’t like having to repeat information in different facilities, and issues such as unstandardized medication histories posed challenges for providers. WellSpan needed a single source of information that all stakeholders could trust to be up to date.
WellSpan’s leaders chose to transition to a single EHR, but they didn’t want to start from scratch and thereby create a huge burden for the clinicians. The IS team began a mass data migration, determined to give the providers clean data.
“We actually did our migrations and go-lives in three phases,” Lackey said. “The first two we did without the benefit of any AI. We used humans, and I tried to help with CCD extracts that were presented for humans to interact with. That helped, but it took a lot of human investment to reconcile that information into the new chart.”
When they saw an opportunity to leverage AI in their process, however, WellSpan became an innovative case study, using support from DrFirst, a Rockville, Md.-based provider of technology solutions and consulting services.
“We were working with a vendor who was giving us our medication-history data, and they had a really cool engine that was already optimizing the medication history. We wondered if we could use that same optimization technology for our CCD extracts of our medications,” Lackey said. “They ended up doing it for us, and it ended up working really, really well.”
Overcoming AI skepticism
The WellSpan leaders understood that many clinicians were skeptical of AI and would need some human assurances.
“We had to do our best to deliver as flawless a process and product as possible, knowing that if we delivered something full of errors, our users would lose faith in it, and that can spell disaster for a project,” Lackey says. “In this project, we did a whole bunch of validation to make sure we were able to catch some errors that the process wasn't able to catch initially. And we corrected for those to make sure that what was being presented to our users was as rock-solid as reasonably possible.”
In the end, 270,000 patent records — including 5.5 million medication records — were run through the AI engine. “Almost 90 percent of the work was optimized, and it saved up to 13 clicks per medication because of all the drop-down (menus) we didn't have to worry about,” Lackey said.
WellSpan thus shaved off five to seven minutes of work per chart, achieved a 60 percent savings over what had been budgeted for the project, and kept their project online during the COVID pandemic. “Our care teams had much better data on day one – the whole process just worked beautifully,” Lackey said.
Having patient charts populated with clean data from go-live led to a better clinician experience and high clinician confidence in the new system. “Being able to instill that confidence early on was helpful,” Gerhart said. “It certainly bred competence and a more successful transition as the clinicians became an integrated part of the organization early on. I think we had much better buy-in from the clinicians, who knew they could be successful with the new solution.”
This experience solidified WellSpan’s belief in AI’s ability to improve the lives of clinicians. It also validated the leaders’ instincts to chase innovation.
“We could have easily just accepted that we were going to do our data migration the same way that we had always done it. I think that's one of the things I would tell other organizations: Never be satisfied,” Gerhart said. “Look at what relationships you have with your vendors, value those, and look for opportunities, when you have them, to leverage something that your vendors are doing for you already in a new and different way.”