Implementing Ambient AI in a Rural Hospital Setting

Julie Demaree, executive director of clinical innovation and transformation at St. Mary’s Hospital in upstate New York, recently spoke with Healthcare Innovation about the implementation there of Suki’s AI Assistant integrated with the hospital’s Meditech EHR.

Many of our interviews about deploying ambient AI assistants are with chief medical information officers in academic medical centers and larger health systems. But small community hospitals are also working with these new digital tools, even if they have fewer resources for support. 

 “I came from another healthcare system where I felt like I had lost the ability to make changes because I couldn’t be nimble anymore,” Demaree said. “And the truth is that here I’m empowered to make decisions, to effect change. I appreciate the opportunity to highlight a small hospital doing great things with fewer resources.”

The clinicians at St. Mary’s have been dealing with a lot of disruption over the past few years, she explained. In addition to coping with the pandemic, in 2020, St. Mary’s left the much larger Ascension health system and became independent, which meant that a lot of support systems, including information technology, became local again. In February 2022, the organization switched EHRs from Meditech Magic to Meditech Expanse, and the physician practices moved from a separate EHR to Meditech. 

St. Mary’s Healthcare is a 130-bed rural hospital northeast of Albany, N.Y., with about 80 employed physicians. Demaree was brought into the organization last year to help optimize the Meditech EHR, including leveraging technology and innovation to decrease the physician documentation burden and decrease burnout.

Healthcare Innovation: Do you have a physician partner you work with on this EHR optimization work?  

Demaree: No, I’m a PA [physicians assistant]. My role is to work with all of our healthcare providers to optimize the system using their feedback. Our CIO and his team manage the hardware, security and infrastructure while the clinical integration team supports the clinical build of the EMR. 

HCI: So when it comes to piloting a voice AI assistant, who is involved in making that decision?

Demaree: When I arrived last July, we started to evaluate different vendors and  brought them in for the physicians to see and evaluate. We also have started a physician steering committee since then.

HCI: When your CEO decided to work on this, was there an obvious problem the organization wanted to address with physicians spending too much time on documentation?

Demaree: Like every small town, we’re trying to recruit physicians, have many physicians nearing retirement and we want to keep them as long as possible? So how do we attract new physicians and encourage seasoned physicians to stay in practice? My CEO and I had worked together previously at a health system where I had implemented ambient listening, so he knew the impact the technology could have, and I was able to use that experience to make decisions this time around. 

HCI: Were there particular things you were looking for?

Demaree: I saw Suki for the first time last September at a Meditech conference. There were a few things that were important to me. I wanted the physicians to see it and see what they liked and what worked. Customer service is really important to me. I didn’t want to have to do all the project management. I needed this to be a seamless integration, and I wanted the physicians to have good support after go-live. 

HCI: Did the work that Suki had done on Meditech integration play into the decision?

Demaree: At the time, Suki was the only AI vendor that had integration with Meditech. As a customer, to have two vendors with a relationship is important. It made it really easy when it was time to actually turn on the API that Suki and Meditech were working together on. Having vendors work together makes a huge difference instead of me being the child in the middle trying to get the parents to talk. 

HCI: Were there things you found in the pilot that needed tweaking to fit your needs? 

Demaree: Network connectivity at some of our sites was insufficient for high-quality data transmission, so we did remediation at some sites. Our users quickly became reliant on the Suki assistnat, and if they are counting on Suki to record the patient encounter and they don’t take any notes, they are really disappointed when it doesn’t work. 

HCI: Do you have any recommendations for other people at small community hospitals who are just starting this process?

Demaree: I think that a demo for physicians is really important. I think  understanding the data that you’ll get back and the customer service that you’ll get is important. And EHR integration is important — what the lift is for that integration for you vs. the vendor. That plays into the cost. This was a really easy implementation. It’s very affordable for us. It’s easy to scale up, and it’s been a very good experience. Of all the things that we can afford to invest in, t this is one that’s paying off on multiple fronts — for patient satisfaction, for physician satisfaction, for quality.

HCI: Do you plan to track the return on investment? 

Demaree: Yes. Our KPIs have been how long it takes for them to complete their documentation, and timely documentation. I can also track the time that they spend outside of their office working in charts. It would be a great benefit if their charges or RVUs continue to go up, which is what we’ve seen initially post-Suki. I attribute this to the AI documentation capturing their  medical decision-making better than previous methods of documentation. Just the satisfaction alone for the physicians and patients is really important. If that is a way for us to attract new physicians and keep physicians here, I think that’s a huge win.