Predicting Influenza Epidemic for United States

Area of Science:

• Epidemiology
• Mathematical Modeling
• Public Health

Background:

• Influenza epidemics cause significant mortality and economic losses globally.
• Accurate prediction of influenza outbreaks is crucial for effective public health interventions.

Purpose of the Study:

• To develop and validate an infectious disease dynamic prediction model for influenza.
• To incorporate transmission characteristics, seasonal effects, and control measure intensity into the model.

Main Methods:

• Proposed the SEIR-CV (Susceptible-Exposed-Infectious-Recovered with Control Variables) model.
• Utilized an adjoint method for inverting critical model parameters.
• Validated the model using 15 weeks of surveillance data for short-term (3-week) influenza prediction in the United States.

Main Results:

• Achieved accurate predictions for the next 3 weeks of influenza epidemics in the US.
• Demonstrated high correlation (r > 0.975) between predicted and surveillance values for predictions spanning 2016-2018.
• Maintained an overall relative prediction error below 10%.

Conclusions:

• The SEIR-CV model is practical and feasible for predicting influenza epidemics.
• The model's ability to integrate multiple factors enhances its predictive power for infectious diseases.