Predicting unobserved exposures from seasonal epidemic data

Area of Science:

• Epidemiology
• Mathematical Modeling
• Stochastic Processes

Background:

• Seasonal fluctuations in contact rates significantly impact disease dynamics.
• Traditional epidemiological models often struggle to capture complex stochastic behaviors.
• Accurate prediction of unobserved disease stages is crucial for public health interventions.

Purpose of the Study:

• To develop a low-dimensional stochastic model for seasonal epidemics.
• To analytically determine the interaction manifold between deterministic and stochastic dynamics.
• To enable data-based prediction of unobserved exposed individuals.

Main Methods:

• Utilized a nonlinear, stochastic projection technique.
• Developed a low-dimensional manifold for model dynamics.
• Employed seasonal Susceptible-Exposed-Infected-Recovered (SEIR) model framework.

Main Results:

• The derived low-dimensional model accurately replicates outbreak timing, amplitude, and phase of recurrent behavior.
• The method successfully captures the interplay between deterministic and stochastic elements.
• Enables long-term prediction of unobserved exposed individuals using infectious data.

Conclusions:

• The nonlinear stochastic projection offers an effective dimensionality reduction for epidemiological models.
• This approach enhances the predictive power of SEIR models with seasonal forcing.
• Facilitates improved understanding and management of seasonal infectious diseases.