In this paper we introduce a model for the spread of COVID-19 which takes into account competing SARS-CoV-2 mutations as well as the possibility of reinfection due to fading of vaccine protection. Our primary focus is to describe the impact of the B.1.617.2 (Delta) and B.1.1.529 (Omicron) variants on the state of Hawai‘i and to illustrate how the model performed during the pandemic, both in terms of accuracy, and as a resource for the government and media. Studying the effect of the pandemic on the Hawaiian archipelago is of notable interest because, as an isolated environment, its unique geography affords partially controlled travel to and from the state. We highlight the modeling efforts of the Hawai‘i Pandemic Applied Modeling Work Group (HiPAM) which used the model presented here, and we detail the model fitting and forecasting for the periods from July 2021 to October 2021 (Delta surge) and from November 2021 to April 2022 (Omicron surge). Our results illustrate that the model was both accurate when the forecasts were built on assumptions that held true, and was inaccurate when the public response to the forecasts was to enforce safety measures that invalidated the assumptions in the model.

The ongoing COVID-19 pandemic highlights the essential role of mathematical models in understanding the spread of the virus along with a quantifiable and science-based prediction of the impact of various mitigation measures. Numerous types of models have been employed with various levels of success. This leads to the question of what kind of a mathematical model is most appropriate for a given situation. We consider two widely used types of models: equation-based models (such as standard compartmental epidemiological models) and agent-based models. We assess their performance by modeling the spread of COVID-19 on the Hawaiian island of Oahu under different scenarios. We show that when it comes to information crucial to decision making, both models produce very similar results. At the same time, the two types of models exhibit very different characteristics when considering their computational and conceptual complexity. Consequently, we conclude that choosing the model should be mostly guided by available computational and human resources.