Cornell has been both applauded and criticized for its fall plan to reopen campus that hinged on successfully implementing the work of University researchers and engineers.
Now, to prepare for Cornell’s spring 2021 reopening, Prof. Peter Frazier, operations research and information engineering, and his team reviewed COVID-19 transmission data from the fall to develop a supplemental testing program for employees and a stricter travel policy for students, as personal travel plans are now only approved in extenuating circumstances.
Frazier also said some groups of undergraduates, including athletes and members of Greek life, are now being tested three times per week, partially because of the more contagious COVID-19 variants that Cornell has identified on campus.
Frazier cites faster-spreading variants, coupled with pandemic fatigue, vaccine hesitancy and limited access to vaccines, as reasons to remain cautious about coronavirus transmission during the spring semester.
These concerns come as on-campus positive case numbers have stabilized following a cluster of cases tied to Greek life that pushed the University to a yellow alert level about two weeks ago.
Despite these concerns, Frazier remains “cautiously optimistic” that cases will remain stable on Cornell’s campus throughout the spring semester.
Leaderboard 2
“Ultimately, we should be very hopeful. Things are getting better,” Frazier said. “Data can lead the way and help support good decision-making.”
Since the summer, the team of professors and students tasked with crafting Cornell’s reopening model reflected on how they used operations research and information engineering to minimize COVID-19 transmission to 351 total cases in the fall.
According to Prof. Mark Lewis, director of the School of Operations Research and Information Engineering, Cornell administrators heavily relied on the large-scale simulation modeling and data analysis involved in operations research to determine whether in-person instruction would be possible at Cornell.
Newsletter Signup
Frazier, who led the operations research team behind the reopening model, explained that college campuses across the nation struggled with extensive COVID-19 transmission, especially due to the large amount of social contact and high rate of asymptomatic COVID-19 infections among college students.
But according to Frazier, a robust system of asymptomatic surveillance testing set Cornell apart by keeping COVID-19 cases relatively low among students and staff.
“There were other universities that were deciding to remain open, but they didn’t give solid justification for doing that. It just seemed like they were ignoring the risk,” Frazier said. “So we thought that maybe because we have this testing program … we could actually do this.”
Frazier added that the University greatly benefited from the veterinary school’s significant testing capacity, Cornell’s close collaboration with Cayuga Medical Center and students’ overall compliance to social distancing and mask-wearing.
However, Frazier recognized that there was a large degree of uncertainty in the model because of the lack of available data on COVID-19 transmission before the start of the fall term. After simulating several worst-case scenarios, Frazier said he and his team were pleasantly surprised that the reality was more optimistic than expected.
“The nature of exponential growth [of epidemics] makes it really, really hard to have an accurate prediction,” Frazier said. “Fortunately, the goal of [our models] was not to have an accurate prediction — it was to [inform] decisions.”
Beyond a rigorous virus-testing protocol, many more engineering decisions shaped Cornell’s reopening plan — including reimagining classroom spaces to accommodate COVID safety precautions.
According to Prof. David Shmoys, operations research and information engineering, the team in charge of scheduling University classes faced a classroom capacity shortage, requiring a tool that could analyze the architecture of a room and assign seating locations for any classroom across campus.
This tool used graph optimization — a computer science technique that can model seats and the spaces between them — by analyzing University architecture drawings maximizing the number of students that could be seated in any given classroom.
The result, according to Shmoys, was a “happy ending.” The red stickers designating where masked students can sit during in-person lectures contributed to no evidence of classroom transmission during the fall semester.
Prof. Oktay Gunluk, operations research and information programming, added that the team’s modeling approaches were slightly too conservative, overestimating the degree of on-campus COVID-19 transmission. However, Gunluk expressed relief that they were prepared to deal with a higher virus prevalence than Cornell actually encountered.
“In reality, we did over-engineer the whole approach,” Gunluk said. “The models and solution techniques we built were like building a Concorde airplane, but at the end we realized we only had to fly 10 miles.”
Although COVID-19 transmission was contained during the fall, the full picture was far from rosy.
Anders Wikum ’21, an undergraduate on the operations research team, said that certain changes made to reduce infection — like cutting out breaks — took a substantial toll on students’ mental health.
“A lot of these changes were necessary, but they weren’t exactly conducive to a learning environment that is relaxing,” Wikum said.
But Wikum said that he felt hopeful that Cornell would prioritize mental health looking to the future, and felt reassured by emails from administrators conveying their commitment to addressing students’ concerns.
Ultimately, Frazier reflected that the operations research behind Cornell’s reopening model helped reduce COVID-19 transmission because Cornell’s top administrators were receptive to the model. As science researchers themselves, President Martha Pollack and Provost Michael Kotlikoff analyzed the evidence presented to them, lending credibility to their decision to reopen Cornell.
“They really understand the work, and so that helps them to trust it,” Frazier said. “Because they’re scientists, they have experience making decisions based on evidence … so those are important factors in allowing our work to have had impact.”