Date of Completion

Spring 5-12-2021

Thesis Advisor(s)

Caiwen Ding; Xiang "Peter" Chen

Honors Major

Computer Science


Computational Engineering


COVID-19 has immensely impacted life as we know it, as the virus quickly spread throughout the entire world in a matter of weeks since its emergence. It has toppled economies, tested healthcare systems worldwide, and has un- fortunately taken the lives of many in the process. While extensive research has analyzed the issue on a large scale, focusing on entire countries and states, there has not been as much focus on the meso-scale, mainly compris- ing towns and cities, due to the lack of available COVID-19 data at this scale. However, in the case of countries like the United States in which interna- tional and interstate travel restrictions were placed, there is reason to believe that many of the coronavirus transmissions would occur at a smaller scale within communities. Thus, using a dataset containing the daily case counts of each town across the state of Connecticut since early 2020, we propose several deep learning models utilizing various neural networks to make case count predictions of each town in the state. One overlooked factor in existing epidemic models is the interaction between towns. We therefore incorporate the distance between the towns in one of our models to measure the strength of inter-town interactions to aid in making predictions. Through our models and their respective prediction results, we aim to provide insights on being better prepared for future pandemics, especially at the community level.