Digital twin technology helps reduce ED wait times

Erie Shores HealthCare’s Emergency Department in Leamington, Ontario, 50 kilometres southeast of Windsor, has reduced the average time until initial physician assessment (PIA) by more than 40 percent – from 7.7 hours to 4.5 hours – thanks to innovative technology from Ontario-based SiMLQ.

The digital system leverages simulation, machine learning and queuing theory to assess process performance measures and optimize department operations.

Co-founded by a trio of Toronto-based academics – Opher Baron, Arik Senderovich, and Dmitry Krass – SiMLQ uses detailed data from the Erie Shores hospital information system to simulate the department’s operations and predict what will happen if specific changes are made.

Normally, explained Baron, a University of Toronto Distinguished Professor of Operations Management at the Rotman School of Management, “it would take a long time to see the effect of a change [that’s made independently by ED staff]. Using our technology, we can do it in four or five seconds. We don’t need to wait to see the effect because we have a digital twin, a live model of the current system.”

Erie Shores’ ED wait times were hovering around five-and-a-half to six hours but had reached nearly eight hours in February 2025, said chief quality officer Dr. Mason Leschyna. These waits were nowhere near the much longer wait times in some hospitals across the country, “but we wanted to improve our performance and funding through Ontario’s Pay for Results program.”

Ranking in the top half of Ontario EDs, “but closer to the middle than the top,” Erie Shores reports annual ED visits of approximately 35,000. However, in recent years it has undergone a transition from a small community focused ED to a medium-sized emergency department serving a larger catchment area with higher-acuity patients.

The change was made to relieve pressure on Windsor Regional’s ED and Essex-Windsor EMS paramedics, who were stuck waiting in hallways for up to 10 or 12 hours when the department was backed up. As a result, Erie Shores ED is now serving about a third of the patients seeking emergency care in the Windsor-Essex County area.

“With that, we’re having sicker patients coming in with more complex presentations and fewer patients with simple concerns,” said Dr. Leschyna. “So, it has become a more demanding place to work and that had contributed to a drain on morale. At the same time, our performance dropped because when you have more complex patients, the work is a lot more difficult and time consuming.

“Instead of seeing a 25-year-old with a cough and cold, we end up seeing an 80-year-old who’s in a nursing home with numerous comorbidities presenting with multiple concerns.”

There was a little less impact on staff who had experience working in busier EDs, but it was a big change for doctors and nurses who were used to the slower pace of the ED prior to the transition, said Dr. Leschyna.

“Clinical administration had already been implementing creative approaches to improve access and flow in the department,” remarked SiMLQ’s Baron. “However, they needed our technology to amplify these ideas and create the capacity to step back from day-to-day operational demands and redesign processes in a more sustainable, effective way.”

“What we aim to do,” added Alwin Hartawan, SiMLQ’s director of business development, “is take that burden off their shoulders and let them focus on patient care. If we can automate this kind of decision-making for them, they can concentrate on saving lives.”

Efforts to improve performance prior to the engagement with SiMLQ were challenging because “they weren’t controlled experiments where you change one variable and everything else stays the same,” said Dr. Leschyna. “As you’re changing, the entire environment is changing around you, so it’s hard to know if what you’re doing is making a difference.”

Another challenge with making changes in the absence of SiMLQ’s technology, noted Dr. Leschyna, is that people in the department have different ideas about how to improve performance, “and without solid evidence to support one over another, it often comes down to whoever presents the most convincing argument rather than a true evidence-based decision.”

The anonymized data that’s used in SiMLQ’s simulation model includes a detailed event log of patient flow through the system from triage to discharge. The model also includes the patient’s age and gender, the zone to which they were routed and the number of staff on the floor.

“The model would see the number of patients flowing through the department and what was happening to each patient,” explained Dr. Leschyna. “They would see the clicks on the computer that represented those tasks. We would give SiMLQ a chunk of that data for a period of time, but we also shared high level process mapping and how we view things as happening from the clinician perspective.”

The biggest change that was made because of the engagement with SiMLQ related to physician scheduling. Instead of multiple doctors starting their shifts at the same time, Erie Shores’ ED moved to a cascading shift schedule with four or five physicians all starting at different times.

“One of the things I suspected but didn’t have evidence for was that wait times depend on when patients arrive during physician shifts,” said Dr. Leschyna.

Patients who arrive at the ED at the start of a physician shift, he noted, are more likely to be seen on an expedited basis, but if they arrive toward the middle or the end of a shift, they’re likely to wait longer because these physicians have already seen a lot of patients and were busy following up rather than seeing new patients. “SiMLQ verified that this was happening,” he said.

With all four or five physicians starting at staggered times, patients are more likely to arrive at the start of a physician’s shift and therefore more likely to have a shorter wait time.

“It made some difference to the average time it takes for a patient to be seen, but where it really matters is in the 90th percentile,” said Dr. Leschyna. “Instead of some patients waiting five hours and other patients waiting 30 minutes, everyone moved down to waiting two or three hours. When you’re looking at the 90th percentile, that makes a huge difference in cutting down those longest wait times, which are so frustrating for patients. Moving the average wait time from three to two and a half hours is important, but moving the 90th percentile down closer to the average is the main thing that patients care about.”

Morale also improved, said Dr. Leschyna, because “no one wants to start a shift with patients who have been waiting seven or eight hours. The first question you’re asked is ‘Why has it taken so long for you to see me?’ It’s much nicer to walk into a shift when patients aren’t frustrated with the long waits.”

Wait times were also reduced by directing more patients to the ED’s lower acuity zone, where patients are assessed much faster than in the high acuity zone. “We suspected this, but SiMLQ brought to light how significant this was.”

“Another change we made was that we became more transparent with physicians about how productive they are and how many patients they’re seeing on a shift,” said Dr. Leschyna. “We created a tracking board showing how much weight doctors are carrying in the department. This encouraged those who were slower to recognize that they were underperforming and adjust their behaviour accordingly.”

While Erie Shores ED staff are pleased with the improvements that have been made, they continue to have access to the model to test additional changes. There is also the possibility of using SiMLQ’s technology to expand the model to cover other areas of the hospital, including diagnostic imaging and the inpatient department, which both impact performance in the ED.