Time-varying reproduction number estimation: fusing compartmental models with generalized additive models.

Area of Science:

• Epidemiology
• Mathematical Biology
• Biostatistics

Background:

• The reproduction number is key to understanding disease spread and control efforts.
• Time-varying factors like behavior, immunity, and viral changes necessitate dynamic estimation of reproduction numbers.
• Existing methods may not adequately address complexities like day-of-the-week effects in surveillance data.

Purpose of the Study:

• To develop a novel, effective, and efficient method for estimating time-varying control and effective reproduction numbers.
• To provide a smoothed measure of the time-varying growth rate from surveillance data.
• To offer a valuable alternative tool for disease control modeling.

Main Methods:

• Utilized a generalized additive model to smooth surveillance data, incorporating day-of-the-week effects.
• Assumed a compartmental model structure to convert smoothed data into estimators for control and effective reproduction numbers.
• Validated the method using both simulated and real-world epidemiological data.

Main Results:

• The generalized additive model effectively smoothed surveillance data and captured time-varying growth rates.
• The derived method provided reliable estimates for time-varying control and effective reproduction numbers.
• The new method demonstrated comparable effectiveness and efficiency to existing tools.

Conclusions:

• The developed method offers an effective and efficient approach for estimating time-varying reproduction numbers.
• This approach is particularly adept at handling day-of-the-week variations in surveillance data.
• The method serves as a valuable addition to the existing toolkit for epidemiological modeling and disease control.