UC San Diego Data Science Undergrads Help Keep K-12 Students COVID-Safe

Scientist examines COVID-19 virus

For over a year of the pandemic, a group of UC San Diego researchers met weekly with epidemiologists at the County of San Diego Health and Human Services Agency (HHSA) to discuss COVID-19 dynamics, analyze populations at higher risk and explore the county’s pandemic response and new ways to mitigate the infection. This early collaboration resulted in several interesting projects, including one led by San Diego Supercomputer Center (SDSC) researcher Ilya Zaslavsky and a team of UC San Diego undergraduate data science students, who developed an agent-based simulation system to assist in COVID-safe school re-openings within San Diego County.

While a variety of high-level policies were considered to make school re-openings as safe as possible, little was known about potential infection spread in a school setting and the efficacy of mitigation measures. At the same time, both teachers and parents were anxious to learn what might happen with children in specific schools, given each site’s variation in design, resources and mitigation plans.

Ilya Zaslavsky
SDSC Researcher Ilya Zaslavsky led the team of undergraduate data science students who developed an agent-based simulation system to assist in COVID-safe school re-openings. Photo courtesy of SDSC External Relations.

Using an early online version of a site simulator, school officials simulated interactions between agents—students and teachers in different grades—as they participated in different types of activities throughout 15 school days: learning at individual desks during class time, group activities, recess time and having lunch in the school cafeteria or classroom. To assess infection risks due to aerosol and droplet transmission, the model used real school floor plans, layouts and capacities of classrooms, cafeterias and recess areas.

Schools were mostly closed for on-campus instruction at the time of these initial simulations, so the modeling team had to rely on data from other countries, from literature and from other fluid dynamics models to generate these initial simulations.

“The simulations allowed us to pinpoint areas in schools that would present higher COVID-19 transmission risks, and to evaluate relative importance of non-pharmaceutical interventions such as wearing masks, reducing class sizes or canceling lunch in the cafeteria and moving it to classrooms,” said Zaslavsky, who is director of the SDSC Spatial Information Systems Laboratory.

According to Zaslavsky, the spatially explicit, agent-based modeling of COVID-19 transmission at schools allowed individual sites and districts to test their plans and match them with their specific spaces, resources and population.

Within several days after the team demonstrated the model in September 2020 to nearly 400 San Diego educators, on a call organized by the County of San Diego, the simulations were run over 300 times. Leslie Ray, senior epidemiologist with the HHSA, said, “At a time when schools were struggling with how to reopen safely, the SDSC team developed an easy-to-use, free tool that allowed schools to test their plans on pixels rather than pupils.”

From the time of the first simulations, the core modeling team, which included UC San Diego data science majors Kaushik Ganapathy, Bailey Man, Johnny Lei and Eric Yu, continued to enhance the model based on the growing literature on COVID-19 epidemiology. Over the year, the team added students and used feedback and advice from the County of San Diego HHSA and from Howard Taras, MD (UC San Diego Pediatrics), key advisor to San Diego schools on managing COVID-19 infection.

From classrooms to school buses

A key concern of school administrators was evaluating risks of COVID-19 transmission on school buses. To extend the classroom model, the transportation division of the San Diego Unified School District (SDUSD) hosted a “data science field trip” to let students take measurements of different types of school buses used in San Diego. They also ran a smoke machine in a moving bus to record air flows at different speeds. Simulating transmission on school buses at the request of SDUSD, the team explored how different seating patterns, bus occupancy, mask wearing and ventilation with windows and hatches open or closed, would affect infection risks.

Women looking at computer screen
The e-Decision Tree was developed by Zaslavsky and Alice Lu (pictured), fourth-year, UC San Diego undergraduate data science student. Photo courtesy of Alice Lu

“In school buses where students with special health care needs cannot always keep their masks on, these models have been very helpful to determine where students should be seated to maximize safety,” said Dr. Taras.

At the same time, using the National Science Foundation’s Extreme Science and Engineering Discovery Environment (XSEDE) with help from the Science Gateways Community Institute, the model components were installed on SDSC’s Comet and Expanse supercomputers and accessed via an Apache Airavata gateway, which allowed the researchers to compute model scenarios much faster and with more input parameters, such as counts of students and staff in each cohort, testing regimens and vaccination rates for students and teachers, mask types and adoption, class occupancy and ventilation characteristics. The system is accessible via GeoACT (Geographically assisted Agent-based model for COVID-19 Transmission).

Making e-decisions

A recent outcome of this collaboration was the e-Decision Tree that is now used by schools around San Diego County to generate instructions to students or school staff who showed COVID-19 symptoms or have been in close contact with infected individuals. The e-Decision Tree was developed by Zaslavsky working with UC San Diego undergraduate data science student Alice Lu. It follows guidance from CDC, the California Department of Public Health, and San Diego County public health orders, and provides schools with an easy-to-use tool to decide when a student or staff should come back to school after isolation or quarantine, and on which dates to get tested. It is not only linked from the San Diego County Office of Education (SDCOE) COVID-19 website, but is also a valued resource on the California Safe Schools for All website.

“By answering a series of five or six brief questions generated by this electronic tool, SDSC has allowed school nurses and primary care pediatricians to replace a time-consuming and complex set of public health regulations with a 30-second survey. And this amazing facilitative device repeats this gift of time several times a day, child after child,” said Dr. Taras.

The team is now working on a Nurse Dashboard that will provide each school with a way to view the status of all students and staff, for the current date and several days in the future and run the e-Decision Tree application for any selected student or staff. It will let school administrators and nurses track how many students should be in school or at home, how many students should get tested, and it will relay individual guidelines to parents. “San Diego County schools are doing a good job of managing and slowing the spread of this virus,” said SDCOE Assistant Incident Commander Bob Mueller. “Given the success of the e-Decision Tree application, which already has been accessed by hundreds of schools in San Diego and neighboring counties, we expect that this new tool will be especially useful.”

This article originally appeared in The UC San Diego News Center, December 9, 2021, and is re-posted with permission in the UC IT Blog.

Leave a Comment

Your email address will not be published.