Understanding and predicting the spread of Omicron

The SARS-CoV-2 Omicron variant spreads more easily than did the earlier variants giving rise to COVID-19. As a result, the upsurge caused by Omicron has surpassed the initial outbreak of COVID-19 that occurred at the beginning of 2020.

Understanding the dynamic evolution and reliably predicting the transmission of Omicron are important in the fight against COVID-19. Cai et al. developed a modified epidemiological model describing and forecasting the spread of this variant.

“This review provides a novel framework for studying the ultra-slow process and predictions of an epidemiological model with limited given data,” said author Changpin Li.

Because Omicron has an incubation period and the reported data usually include daily new infected and removed cases, the susceptible-exposed-infected-removed, or SEIR, model was used in the study.

However, Omicron transmission is more concealed than previous variants, resulting in a slow increase initially. To characterize Omicron’s spread, the researchers also used the Caputo-Hadamard fractional derivative to refine the classical SEIR model. In addition, a data-driven deep learning framework based on physics-informed neural networks was applied to calibrate model parameters.

Simulation and reliable prediction of the resulting model are based on a Shanghai data set. The researchers note that this modeling framework is applicable to other large cities around the world such as Berlin and New York.

“In future work, physics-informed neural networks will be applied to solving inverse problems of a coupled system given by partial differential equations with time-dependent parameters,” said Li.

Source: “Fractional SEIR model and data-driven predictions of COVID-19 dynamics of omicron variant,” by Min Cai, George Em Karniadakis, and Changpin Li, Chaos (2022). The article can be accessed at https://doi.org/10.1063/5.0099450 .