Fuzzy-SIRD model: Forecasting COVID-19 death tolls considering governments intervention

Area of Science:

• Epidemiology
• Computational Modeling
• Fuzzy Logic Systems

Background:

• Compartmental models are crucial for disease trend analysis but lack adaptability to time-varying parameters.
• Existing models struggle to account for dynamic societal responses to interventions.

Purpose of the Study:

• To introduce a novel Fuzzy Susceptible-Infectious-Recovered-Deceased (Fuzzy-SIRD) model addressing limitations of conventional compartmental models.
• To enhance disease modeling by incorporating dynamic government interventions and societal responses.

Main Methods:

• Utilized an interval type 2 Mamdani fuzzy logic system to model government intervention uncertainty.
• Employed a first-order linear system to capture the dynamics of societal response.
• Optimized model parameters using the Particle Swarm Optimization (PSO) algorithm with Root Mean Square Error (RMSE) as the objective function.

Main Results:

• The Fuzzy-SIRD model demonstrated significant improvements in predicting COVID-19 death tolls across seven countries.
• Achieved an average RMSE reduction of 45.83% for short-term and 72.56% for long-term predictions compared to the SIRD model.
• Successfully established a semantic relationship between the basic reproduction number, government intervention, and societal response.

Conclusions:

• The proposed Fuzzy-SIRD model offers a more adaptable and accurate alternative to traditional compartmental models for infectious disease forecasting.
• The model's ability to integrate fuzzy logic and dynamic systems enhances its utility in managing public health crises.
• The findings highlight the importance of considering dynamic interventions and societal reactions in epidemiological modeling.