Estimating the instantaneous reproduction number () by using particle filter.

Area of Science:

• Epidemiology
• Mathematical Modeling
• Public Health

Background:

• Accurate estimation of the effective reproduction number (R0) is crucial for monitoring coronavirus disease 2019 (COVID-19) transmission.
• Existing methods may produce biased R0 estimates by not accounting for real-world factors like confirmation delays, pre-symptomatic transmission, and imperfect data.

Purpose of the Study:

• To develop and evaluate a novel method for estimating R0 that incorporates real-world data complexities.
• To compare the performance of the proposed method against existing approaches for R0 estimation.

Main Methods:

• Expansion of the susceptible-exposed-infectious-recovered (SEIR) model to the SEPIAR model, including pre-symptomatic (P) and asymptomatic (A) states.
• Utilization of both stochastic and deterministic SEPIAR models to generate simulated datasets reflecting real-world data challenges.
• Application of a particle filtering method for R0 estimation and comparison with the EpiEstim approach.

Main Results:

• The particle filtering method accurately estimated R0 even with delayed data, pre-symptomatic transmission, and imperfect observations.
• The particle filtering method demonstrated superior performance based on root mean square error (RMSE), especially with short-term R0 fluctuations and right-truncated data.
• Performance was comparable to EpiEstim when perfectly deconvolved infection time series were available.

Conclusions:

• The SEPIAR model combined with particle filtering provides a robust tool for COVID-19 transmission trend prediction.
• This approach enhances COVID-19 transmission monitoring and aids in evaluating intervention strategies.
• The findings support informed public health policy development for disease control.