Identification of COVID-19 spread mechanisms based on first-wave data, simulation models, and evolutionary...

Area of Science:

• Epidemiology
• Mathematical modeling
• Infectious disease dynamics

Background:

• COVID-19 pandemic highlighted limitations of standard SI, SIR, and SEIR models for predicting disease spread.
• Novel model structures are needed to capture complex, unobserved mechanisms driving COVID-19 transmission.

Purpose of the Study:

• To investigate COVID-19 spread mechanisms by parameterizing and comparing various epidemiological models.
• To identify superior models for accurately fitting first-wave pandemic data.

Main Methods:

• Analysis of COVID-19 data from Our World in Data for the first wave.
• Comparison of SI, SIR, SEIR, SEIUR, and Bass diffusion models using differential evolution optimization (L-SHADE).
• Calculation of reproduction rates (R0) for 61 countries based on best-fit model parameters.

Main Results:

• Modified Bass diffusion and SEIR models demonstrated superior performance in fitting cumulative infection curves compared to classical models.
• Modified SEIR outperformed classical SEIR in 43/61 countries; Bass diffusion outperformed SI in 57 countries.
• Identified limitations of traditional models and highlighted the importance of external factors in transmission dynamics.

Conclusions:

• Modified epidemiological models, particularly those incorporating external spread factors, offer improved accuracy for COVID-19 prediction.
• A significant, non-contact-dependent mechanism influencing COVID-19 spread dynamics was identified.
• These findings provide a basis for refining predictive models for infectious diseases.