Advanced Models for COVID-19 Variant Dynamics and Pandemic Waves

Identifying the driving forces of COVID-19 case counts can help decision makers predict possible effects of virus on populations. This would allow for more swift and directed mitigation tactics, possibly even before new case waves appear. We analyze the role of virus mutations in the dynamics of infection spread via the comparison of cases-over-time data with variant specific data. What we find is a strong correlation between characteristic waves in cases and the evolution of the variant mutations themselves. Namely, when a new variant becomes dominant, it is usually followed by a local maximum in cases. We then use this information to fit an epidemiological model which couples ordinary differential equations with Markov chain dynamics to allow for viral mutation. We see that variants dynamics in such a model is enough to both elicit the characteristic waves in cases, and estimate when they appear over a time horizon. This study pave the way to the use of epidemiological models with variants dynamics for accurate predictions and to guide interventions.