Discrete-event simulation project at Lackey Clinic wins MODSIM first prize
Molly Atwater earned two degrees in five years at William & Mary.
After her May 12 graduation, she has wedding plans to finalize as well as organizing relocation to D.C., where she will take up a job in a consulting firm.
It’s a busy time, but Atwater will need to find a minute to make room on an already impressive resume for an award-winning analysis of the operations of a free clinic based in Yorktown.
Atwater’s analysis of Lackey Clinic won first place in the MODSIM World Conference Challenge Competition, recently held in Norfolk. She graduated from William & Mary in 2017 with a bachelor’s degree in applied mathematics and this year finished a master’s degree in the university’s Computational Operations Research (COR) program.
Rex Kincaid, the director of the program, said COR is an interdisciplinary program, incorporating elements of William & Mary’s departments of mathematics and computer science, applied science and public policy. Kincaid is Chancellor Professor in the Department of Mathematics, while COR is a specialization within the M.S. program in computer science.
“It’s a small, intimate program. We typically have about 10 new students each year,” Kincaid said, adding that COR faculty includes four members with Ph.D.s in operations research, plus an adjunct who also has an operations research Ph.D.
A 30-year history at William & Mary
Operations research as a discipline arose from the military and Kincaid said William & Mary’s math department added the graduate program nearly 40 years ago after an academic reviewer pointed out the number of military installations in the region. Operations research grew beyond military uses and the discipline is now a mainstay in schools with engineering programs.
“What I like about operations research is that theoretical developments are motivated by real-world applications,” Kincaid said.
Atwater’s analysis of the patient flow at Lackey Clinic is a good example of an operations research application. The clinic began operations in 1995, and now offers a broad range of health care services to indigent patients five days a week.
“I love their mission that they are completely free of charge. They’re only available to individuals who fall below a certain percentage of the poverty line,” she said. “They’re really reaching out into the community and making a difference there. I saw I could do some serious work in terms of improving that process.”
The Lackey Clinic had approached Larry Leemis, a professor in the William & Mary Department of Mathematics and a faculty member in the COR program about doing an analysis of the clinic’s operations, which have grown substantially over the years. Leemis suggested that Atwater take on the project.
Leemis said that Atwater’s project is known in the operations research field as a discrete-event simulation. It’s a powerful tool for modeling a series of well-defined events and is also used for systems much more complex than a patient’s progress through a doctor’s appointment. Atwater’s own medical history, chronicled in her MollyOllyOstomy Instagram account, added to her interest in the Lackey Clinic project.
“I’ve had medical issues of my own and I’ve spent enough time waiting for appointments in doctors’ offices to know that it’s not always the most efficient process,” she said.
Not just like a doctor’s office
Atwater explained that the Lackey Clinic faces circumstances that are different from most physicians’ offices. For one thing, the clinic has a staff that is a mixture of paid and volunteer personnel.
“There are two full-time doctors and the rest are volunteers,” she said. “Same with the nurses — there are a couple of people full time and the rest are volunteers.”
The other difference is in the patients and Atwater said the patient census was even more significant for her analysis than the clinic’s mix of paid and volunteer health care staff.
“Where it gets interesting is that more than 70 percent of these patients have more than three chronic conditions,” she said. “So, you’re seeing a patient who has, for example, diabetes and heart disease and joint problems. ”
Such a patient census means that patient appointments tend to be lengthy, Atwater explained. She began her work by spending five or six days at the clinic, talking with doctors and nurses and, as she says, “just trying to get a sense of what a day in the life of Lackey looks like.”
She found that a day in the operational life of Lackey Clinic to be pretty familiar: Patients check in at the front desk. A nurse takes them from the waiting area for triage — blood pressure, temperature, heart rate, weight. Then the patient goes to the examining room to await the physician.
Converting it all into math
After observing the patient flow and sitting and talking with the doctors and nurses, Atwater said her next task was to “try to convert their words into math.” She used SIMAN, a simulation programming language, to assign variables to each aspect of a day in the Lackey Clinic, creating a computer simulation of that day. Her simulation revealed a number of interesting things. For one, the people at Lackey were not slacking.
“One doctor who worked on a Wednesday had a utilization measure of about 90 percent,” Atwater said. “So, 90 percent of the time, she was doing something patient-related, which gives very little wiggle room if a patient appointment runs a little bit long.”
It was clear that bits and pieces of time were adding up. Atwater found that some patients were waiting to see the doctor for periods as long as 45 to 50 minutes. Analysis of the model she built revealed a couple of things that she recommended to the clinic board.
“Adding another doctor, and reallocating some of the nursing staff so that two nurses were doing triage duty instead of one, significantly decreased the longest wait time down to like 15 minutes,” she said.
Atwater’s analysis also revealed an easily remedied redundancy. Nurses doing triage were entering data on a paper form on a clipboard, then transposing the information into a computer in another room. She and clinic staff discussed the time expended in entering data twice, then Atwater recommended a change in procedure.
“The idea is that the nurses could have a tablet or some sort of in-room computer and they could input the information right there while they’re with the patient,” she explained. “It would save between two and five minutes, which doesn’t sound like a lot, but if you’re seeing multiple patients, it could add up to another appointment time each day.”
The project was a big hit all the way around
Atwater submitted an analysis paper on the Lackey workflow to the MODSIM World competition. She was one of three finalists and MODSIM invited her to present her project at the April conference. It went over well.
“The judges loved the approach. I got lots of compliments on my presentation,” she said. “That’s one thing that the program here does really well. You give a lot of presentations, so you get comfortable in front of a crowd.”
She received specific glowing reviews on a number of aspects of both her presentation and the analysis itself:
“Having judged this event several times now, your presentation and research approach were, by far, the best I have seen,” one judge wrote. Another judge commented: “The amount of detail provided throughout was excellent. It allows for the reader to gain a full understanding of the background leading up to the simulation, as well as the data results being presented. At no point was there a need to re-read any portion of the paper due to misunderstanding or loss of logical flow of the presentation. Very well done!”
More importantly, the people at the clinic were equally impressed. Atwater presented her recommendation to the Lackey clinical staff and again to a meeting of the facility’s governing board. She said the Lackey Clinic staff were enthusiastic about her analysis and her recommendations.
“I brought the mathematical reasoning behind it,” she said, adding that her model worked hand-in-hand with her own human insight. “The computer won’t tell you they need to get tablets; the computer doesn’t know that. The computer only knows the model. You know the people behind it. You know how the culture works, so you can make better suggestions.”
