We have conducted one SIR model and three agent-based modelings to predict and address the COVID-19 spreading at the UConn campus. We extend the SIR model to a spatial dimension, using RStudio, QGIS, and NetLogo software. Results present that phase 2 of the reopening policy is the most effective way to decrease virus transmission. The spatially-explicit models may help policy-makers for targeted interventions and targeted reopening business in specific locations. Areas free from infection can be free of restrictions if it is deemed safe to do so. The spatially-explicit simulation results would make it easier for decision-makers to locate the hotspots of covid-19 and take proper intervention actions in these hotspots areas while open business in other less affected areas.

In this paper, we demonstrate that phase 2 of the reopening policy can minimize the virus's spreading while ensuring the regular operation of campus and businesses. 

This paper analyses the risk level of different policies, which are commonly used in public health services on campus, to give suggestions to campus policymakers with evidence about what can be minimized the spread and safely quarantined when students are tested positive. We show that the prediction of the SIR model may be a useful input into campus policy, which evaluates which buildings are safe to allow quarantine and which have potential risks. 

More detailed data is needed for improving the accuracy of models. Since COVID-19 has much uncertainty, we still need to collect the data and modify our models to fit the real situation better. Using the UConn data instead of Mansfield data would significantly improve the models' accuracy and help us have a better understanding of the models. We do not need to wait for the next outbreak before building and analyzing the model.

Currently, UConn is doing an excellent job of providing hand sanitizer, wearing masks, and quarantine buildings. The policies including quarantined students' need to eat during design hours, providing medical needs and free testing, and residential buildings are quarantined once there are positive cases.

Still, the policies right now are not sufficient to address the rising number of infections. Accordingly, as policymakers move forward with plans for further steps, it is essential that they focus on providing adequate and sufficient tests while safely reopening campus in order to prevent the spread of disease. Students who go back home during the semester need to get tested before going back to dorms or classrooms. Also, people who are not living on campus cannot go to dorms or classrooms without permission. 

Isolate people confirmed positive cases to apartments such as Northwood apartment, Charter Oak apartment, or Busby suites. As confirmed cases increase, it will be significant to stop further spreading, especially in residential halls. The population density in residential halls is different from houses. Students' interactions in the building before quarantine are high, such as elevators, bathrooms, and lounges. Once there is a confirmed case, other students who live in the same building have higher chances of getting infected. Currently, the residential policy quarantines the whole building, and students cannot attend in-person classes [7]. However, it only decreases the exposure in classrooms, but still has risks to infect in the residential halls. Students who confirmed positive should be isolated somewhere with a lower population density and far from the central campus.  

As the vaccine for COVID-19 coming available, UConn needs to make a list of the allocation priority. Minority groups such as International students and out-of-state students are more vulnerable because they are far from home and having less access to local resources. Compared to long-term residents, who live on campus, transfer students and off-campus students have higher risks of carrying viruses. We can easily trace whom the students who live on campus have interaction with. In contrast, it is hard to track who had contact with the student who transferred or live off-campus. Before the new semester begins, students and staff who come to campus need screening and testing. 

There are limitations to our paper. First, due to there is no public data on UConn students' confirmed cases, we estimate the cases at UConn by the proportion of Mansfield's population and the number of infected people. Second, the age group of UConn is mainly young people, which is different from Mansfield. Thus, the recovery rate at UConn may be higher compared to Mansfield. Because of the data uncertainty issue, the models fitted to covid-19 confirmed case data in this study are probably not very reliable. Model parameters associated with covid-19 transmission rate probably also have uncertainty. The parameter uncertainty may propagate through the models, therefore, unavoidably generated uncertainty in the simulation results. Despite these limitations, we believe that this paper helps arrange the reopening policy, providing risk scores of buildings and shops, especially when the holiday is approaching. There are many factors related to covid-19 transmission such as from political and societal issues to ethical and cultural standards, which are difficult or impossible to be represented in any model. It is impossible for any model to predict the situations of the covid-19 epidemic very accurately. Our simulated results can only serve as a guiding tool for policy-makers. We may incorporate more factors to better simulate the complex covid-19 situation. However, we also realize that the increased complexity of a model usually comes with increased difficulty for manipulation, analysis, computation, and implementation. The complexity of a model does not mean it could increase the accuracy of the simulation results.

1. The Flu Season. (2018, July 12). Retrieved October 24, 2020, from https://www.cdc.gov/flu/about/season/flu-season.htm

2. Robson, D. (2015, October 18). The real reason germs spread in the winter. Retrieved October 24, 2020, from https://www.bbc.com/future/article/20151016-the-real-reason-germs-spread-in-the-winter

3. Benjamin Rader, MPH, Christina M Astley, MD, ScD, Karla Therese L Sy, MS, Kara Sewalk, MPH, Yulin Hswen, ScD, MPH, John S Brownstein, PhD, Moritz U G Kraemer, DPhil, Geographic access to United States SARS-CoV-2 testing sites highlights healthcare disparities and may bias transmission estimates, Journal of Travel Medicine, , taaa076, https://doi.org/10.1093/jtm/taaa076

4. Satsuma, J., Willox, R., Ramani, A., Grammaticos, B., & Carstea, A. (2004, January 24). Extending the SIR epidemic model. Retrieved October 27, 2020, from https://www.sciencedirect.com/science/article/pii/S0378437103012287?casa_token=AtYHM_q6Ca8AAAAA%3AmV4r9_ZTsQPeLX9_2fNxmu8kgQU5_U3CmrPauCUNYNSLS_e7IgfMS_0cvaH8eoDyT33EQxvXoEIM

5. Bonabeau, E. (2002, May 14). Agent-based modeling: Methods and techniques for simulating human systems. Retrieved October 29, 2020, from https://www.pnas.org/content/99/suppl_3/7280

6. CT.gov. (2020). Sector Rules for Reopen. Retrieved October 29, 2020, from https://portal.ct.gov/DECD/Content/Coronavirus-Business-Recovery/Sector-Rules-and-Certification-for-Reopen

7. Hubbell, D. (2019, December 23). Novel Coronavirus Disease (COVID-19) Information. Retrieved November 17, 2020, from https://studenthealth.uconn.edu/updates-events/coronavirus/