Peterborough Regional Health makes AI push, rolls out Peregrine

LAS VEGAS – At a HIMSS session on how hospitals are deploying real-world AI solutions, Evan Lyons, executive vice president and chief information officer at Peterborough Regional Health Centre (PRHC), described a major AI-driven platform that’s about to be rolled out at his organization in the next few months.

PRHC is a community hospital located 140km east of Toronto, serving a broad urban and rural referral population of 600,000. It’s unusual for a hospital of its size, outside of a major urban centre, to have orchestrated such a large-scale AI program.

Lyons spoke at the annual HIMSS conference in March.

He said the hospital has integrated its main sources of data and linked them to an AI-enabled data platform, so that clinicians and administrators can find the answers to difficult questions faster than ever before. The system is designed to help them make more informed decisions for both clinical and administrative work.

Called Peregrine, the AI-powered platform makes use of underlying technologies from Microsoft, including Fabric for integration and Copilot for generative AI. “We’ve now integrated over 18 different data sources, and we’ve flowed them into the platform,” Lyons said.

Key to the effort has been the leadership of PRHC president and CEO, Dr. Lynn Mikula, who has championed new ways to handle rising costs and the shortage of staff in healthcare today.

She has been a huge supporter of innovative technologies like interoperability solutions, business intelligence and AI to enable staff to make better decisions and to work more effectively.

As she noted in a press release issued by PRHC, “It’s difficult to overstate the effects Peregrine will have on the way we do things at the hospital. Having real-time, curated data at our fingertips through a custom-built platform that has been designed to inform and support the decisions we make – this is a vitally important tool when it comes to the lifesaving work we do here every day.”

PRHC is encouraging people to query the system to find their own answers to difficult questions.

Lyons said one of the more technically inclined physicians worked with the AI system on departmental staffing.

“They took the data and did some modeling with it. In a department where they’ve done scheduling the same way for about 15 years, they’re now changing their scheduling system, based on the data,” said Lyons.

Peregrine is expected to become very helpful in case costing, too, as departments try to keep spending under control.

The system allows users to model different scenarios, looking at which solutions are most efficient, and which are delivering the best results at the lowest cost.

“It doesn’t help to tell clinicians, ‘do surgery more cheaply or see patients faster,’” noted Lyons. “They don’t want to hear that.”

“But if we know that a particular product can be replaced at a better price while maintaining or improving clinical outcomes, and it’s being used by our peers, there’s a much different conversation that takes place.”

In the orthopedics department, Lyons said, use of Peregrine has already helped clinicians focus on the most cost-effective solutions. “We used to have seven different knee systems for implants. Through the case-costing work we’ve been doing, that’s come down to three.”

With many of its systems now integrated, and layers of business intelligence and generative AI put into place, the hospital is soon going to launch its own command centre, making use of the connected data to monitor patient flow throughout the hospital.

That will not only enable staff to monitor logjams and problem areas in the hospital, but it will also give them predictive abilities.

For example, they might be alerted that a wave of patients is expected the next day, with a certain level of acuity, enabling the team to get ready for them.

Bed management is, moreover, something that can be greatly improved by artificial intelligence. “You might have 20 people who need beds, and 18 beds are immediately available,” said Lyons. “The system might be able to tell you that a certain bed is available right away, but if you wait 20 minutes, a bed will become available in an area of the hospital that will better serve that patient’s needs.”

Significantly, PRHC is seeking to work with its four partner hospitals in the region, extending the Peregrine platform to them so they can obtain a region-wide view of patient patterns and logjams.

“That could be a big win for all of us – and not just within our region, but well beyond it,” said Lyons. “We can exchange information securely, and we can co-develop solutions that help us collectively. It means we won’t be repeating the same hires and rebuilding the same infrastructure.”

Also in the works is a plan to extend the benefits of AI to patients. Lyons said AI agents for patients are in development, enabling them to ask questions about their own healthcare.

For example, he said after a visit to the fracture clinic for a broken bone, a patient is typically discharged with instructions for the next six months. He or she may quickly forget what was said.

But the agent will enable them to ask questions in natural language and obtain answers – at any time. Moreover, the agent will be linked to the patient’s own health record, so that it can tailor answers specifically to that person.

The same agent can also pre-screen patients in the clinic. “There might be indications of intimate partner violence,” said Lyons. “Then we might approach the waiting room differently, taking the patient to a private screening room so we can have that conversation.”

He said the AI agent can also flag warnings for things like osteoporosis, alerting clinicians to order tests and providing information and instructions to the patient about the disease.

Lyons told the audience at HIMSS that the driver of technological adoption at PRHC is the fast-growing demand for healthcare services. “We have increasing volumes, increasing complexity in the patients we’re seeing. And we don’t have the budget to infinitely expand our workforce.”

“So, we really started looking at how can we do things differently,” he added. “How can we do things better? We sat down as a leadership group and looked at opportunities that we might capitalize on to allow our workforce to get more mileage.”

Lyons said that brainstorming highlighted issues like: “What parts of their work are they doing that we don’t want them doing? What other work could they be doing that would be more aligned to our core values and mandate, providing excellent patient care?”

“So, we looked at… a few different lenses, but specifically on the sort of health operations side, what are we doing right now where someone spends a day scouring an Excel spreadsheet looking for insight? Where is the person who’s not getting any insight at all, there’s nothing being delivered to them? And who is spending all of their time doing the mundane repeated tasks?”

One of the goals was to reduce these mundane tasks, handing them off to technologies, so that employees could focus more on patient care.

“We wanted to put tools in their hands that allowed them to focus on bringing the joy back into their work, that let them bring empathy into patient care,” said Lyons.

AI was immediately discussed as a possible solution, but the team at PRHC quickly realized the workforce wasn’t ready for artificial intelligence – emotionally or technologically.

Creating the dashboards and business intelligence system served as springboard into AI.

PRHC brought in these technologies and educated staff and clinicians, encouraging them to familiarize themselves with them.

“We put dynamic dashboards into people’s hands for the first time. That’s a jump,” said Lyons.

“If we had gone straight to AI, we would have lost a lot of people … there’s already enough distrust of the technology in the space. So, we started fairly simple.”

Once people became more familiar with the dashboards, the team at PRHC added Microsoft Copilot on top of Power BI, allowing them to ask natural language questions of the dashboards and business intelligence system.

The organization also wanted to build the governance structure needed before unleashing AI.

“One of the biggest concerns we had was that we’d have 6,000 agents in our organization doing random things,” asserted Lyons. “We really wanted to be focused on how they actually tie together.”

With these parts of the puzzle in place, the organization then brought in generative AI. It’s embedded into all data systems, so that employees can access the data they need.

“We’re now getting people to realize that these tools are in place to help them,” said Lyons.