An epidemiological modelling approach for COVID-19 via data assimilation

Area of Science:

• Epidemiology
• Computational modeling
• Public health policy

Background:

• The 2019-nCov pandemic necessitates robust policy evaluation to mitigate social and economic impacts.
• Real-time data integration is crucial for accurate forecasting and effective intervention strategies.
• Existing models may lack the granularity needed for precise inference on infection dynamics.

Purpose of the Study:

• To develop and evaluate an epidemiological model for real-time forecasting and policy assessment of the 2019-nCov pandemic.
• To enhance granular inference on infection numbers and disease parameters using a custom SITR model.
• To provide a scalable and extendable modeling framework for future pandemic response.

Main Methods:

• Development of a custom compartmental SITR (Susceptible-Infected-Treated-Recovered) model.
• Application of variational data assimilation for real-time data incorporation.
• Hybrid data assimilation approach to ensure robustness against initial conditions and measurement errors.
• Analysis of infection rates and disease parameters (transmissibility, recovery) in the UK, US, and Italy.

Main Results:

• The SITR model provides granular inference on infection numbers, updated in real-time as new data becomes available.
• The hybrid data assimilation approach enhances model robustness and reliability.
• The model successfully infers key epidemiological parameters, including disease transmissibility and recovery rates.
• Comparative analysis of infection dynamics across the UK, US, and Italy was conducted.

Conclusions:

• The proposed epidemiological model offers a powerful tool for real-time pandemic forecasting and policy evaluation.
• The SITR model's parsimonious and extendable parameterization allows for adaptability to new data and locations.
• This approach enhances preparedness for future pandemics by providing scalable and robust analytical capabilities.