Exploring memory effects: Sparse identification in vector-borne diseases.

Area of Science:

• Epidemiology
• Computational Biology
• Mathematical Modeling

Background:

• Predicting vector-borne disease burden with limited data is challenging due to complex transmission dynamics.
• Vector ecology and human behavior introduce nonlinearities and delays in disease spread.

Purpose of the Study:

• To develop a data-driven framework for discovering disease transmission mechanisms from time series data.
• To forecast vector-borne disease risk using limited surveillance information.

Main Methods:

• Extended Sparse Identification of Nonlinear Dynamics (SIDyN) to systems with distributed memory.
• Applied the framework to severe fever with thrombocytopenia syndrome (SFTS) using human incidence and temperature data.
• Integrated mechanistic covariates to assess model robustness and forecasting improvements.

Main Results:

• The data-driven model successfully identified key features of tick-borne disease dynamics.
• Forecasting accuracy was strong using only incidence and temperature data, with no significant improvement from mechanistic covariates.
• The framework demonstrated robustness and yielded interpretable integral representations for epidemiological forecasting.

Conclusions:

• The proposed data-driven framework offers a scalable strategy for forecasting vector-borne disease risk.
• This methodology can inform public health decision-making, especially under data limitations.
• The approach prioritizes predictive accuracy while providing interpretable models for disease dynamics.