This application displays the results of an agent-based model (ABM) developed to simulate the transmission of SARS-CoV-2 among students, faculty, and staff at the University of California San Diego (UC San Diego).

The model was used to inform the UC San Diego Return to Learn program, which aims to prevent the spread of SARS-CoV-2 on campus through three adaptive pillars that include transmission-reduction strategies, monitoring viral activity, and public health interventions. The model incorporates important features related to risk at UC San Diego, such as community composition (staff, faculty, and students on or off campus), campus residential configuration, and course registration. The model includes a constant risk of importation of virus from the outside community.

The ABM results presented focus on assessing the potential impact of two strategies to reduce risk of SARS-CoV-2 (the virus that causes COVID-19) transmission: (1) masking and social distancing (0 to 75% adherence) and (2) asymptomatic testing (monthly to twice weekly testing). The simulations predict outbreak sizes associated with each viral introduction, cumulative infections over the course of an 80 day term, and peak isolation housing need across a term.

The model was developed by researchers at Mathematica and UC San Diego and will be adapted and refined as further data are collected. For more information on UC San Diego’s Return to Learn program visit here.

Model Overview

Agent-based models (ABMs) are computational models that imitate how interactions of individuals (“agents”) contribute to community-level outcomes. Over the past 50 years, ABMs have been used to evaluate the effectiveness of efforts to control spread of disease, reduce teenage smoking, diffuse technology, and update agricultural policies. The University of California San Diego (UC San Diego) uses this ABM to advise data-driven policy decisions surrounding its “Return to Learn” program.

The ABM simulates an individual's progression through seven stages:
  1. COVID-19 negative
  2. COVID-19 positive, pre-symptomatic
  3. COVID-19 positive, asymptomatic
  4. COVID-19 positive, symptomatic
  5. Hospitalization
  6. Recovery
  7. Death
Each day, an individual either remains in the current stage or transitions to another stage. These stages are assumed in the background but the visualization accounts for the hospitalization at UCSD.

The ABM analyzes the following interactions when considering the social distancing component:
  1. Class - interactions within the classroom
  2. Residence - interactions with roommates, suite-mates, and building-mates
  3. Campus - interactions on campus outside of classes and residences
  4. Community - interactions occurring from social interactions outside the university

Mitigation strategies

The ABM simulates the transmission of SARS-CoV-2 among students, faculty, and staff at the UC San Diego under varying levels of adherence to masking and social distancing (0% to 75%) and testing frequency (monthly to twice weekly).

An effective strategy (in terms of infection reduction) would limit within-university transmissions, increasing the proportion of all transmissions that occur outside of (and thus out of the control of) the school.

Masking and social distancing
We investigated the potential impact of students adhering to masking and social distance guidelines. We varied the level of adherence to masking and social distancing from no interaction adhere to these precautions (0%) to three quarters of interactions occur with precautions (75%).

In the model, all residential students will be tested two weeks prior to the start of classes and again right before the start of classes. If a student is positive, they would be placed in isolation housing. After the start of classes, individuals with in-person classes or reside on-campus (referred to as campus testing) will be tested at differential rates compared to individuals with no in-person classes or a non-resident (referred to as non-campus testing). We investigated on-campus testing rates of monthly, every 2 weeks, every week, 2x weekly. We assumed that non-campus testing rates will by monthly.


The ABM is based on underlying assumptions about several key parameters, including rates of transmission in different contexts and susceptibility probabilities. We have done our best to base these parameters on estimates from the emerging SARS-CoV-2 literature, but there is still scarcity in information regarding the spread of SARS-CoV-2, especially for universities. Further, there is significant uncertainty in identified estimates for transmission and susceptibility rates. For parameters where we could not find reliable estimates in the literature, we used rates that seemed plausible relative to other rates in used in the model.

UC San Diego’s Return to Learn program aims to prevent the spread of SARS-CoV-2 on campus and promote safety this fall through three adaptive pillars that include transmission-reduction strategies, monitoring viral activity and public health interventions.

Mathematica researchers have developed and used ABMs that simulate the spread of SARS-CoV-2 for K-12 schools (Gill et al, 2020a, Gill et al, 2020b) and universities to help them make data-driven decisions tailored to their communities about re-opening and operating policies. If you would like to learn more about using modeling to develop data-driven decisions for potential mitigation strategies, please contact Andrew Hurwitz,

The New York Times provides and interactive map showing COVID-19 case counts at U.S. colleges and universities since the start of the pandemic.