Integrated disease model considering mutation-induced infection waves with COVID-19 cases.

Area of Science:

• Epidemiology
• Mathematical Modeling
• Public Health

Background:

• COVID-19 pandemic presents complex dynamics due to successive waves and emerging variants.
• Traditional Susceptible-Infected-Recovered (SIR) models struggle to accurately capture these evolving dynamics.
• Need for adaptive epidemiological models that incorporate variant-specific data.

Purpose of the Study:

• To develop and validate an integrated empirical model for COVID-19 that accounts for dominant variants.
• To provide a data-driven framework for recalibrating models as new variants emerge.
• To establish theoretical justification for combining variant-specific models.

Main Methods:

• Integrated multiple sigmoidal (logistic) curves, each representing a dominant variant's cumulative infections.
• Utilized real-world data from Our World in Data (cases) and GISAID (variant prevalence).
• Employed the Pruned Exact Linear Time (PELT) algorithm to determine when models could be summed (approx. 50% variant dominance).

Main Results:

• The integrated model demonstrated markedly improved accuracy compared to single-strain approaches.
• Model performance was evaluated using Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE).
• Findings were validated using data from fourteen countries and global aggregates.

Conclusions:

• Iteratively updating epidemiological models with emerging variant data enhances predictive performance.
• The integrated framework offers a feasible and advantageous approach for pandemic management.
• The study provides actionable insights for ongoing and future pandemic response strategies.