Process-guided machine learning model for forecasting Ae. albopictus population dynamics across Europe.

The epidemiology of mosquito-borne diseases in Europe is rapidly changing due to climate change, urbanization, and globalization of travel and trade. The invasive mosquito Aedes albopictus has expanded rapidly across Europe. This expansion has been linked to recent increases in autochthonous dengue and chikungunya transmission.

Given the species complex ecology, replicating its population dynamics through ecological model remains challenging because traditional approaches based solely on statistical data-driven or mechanistic approaches may have inherent limitations.

Here, we present a Process-Guided Machine Learning (PGML) framework that integrates mechanistic mosquito population estimates with machine learning models. Mechanistic outputs are used as biologically meaningful predictors alongside environmental and entomological data to estimate spatially explicit mosquito occurrence.

The PGML framework captures complex environmental interactions while reducing unrealistic extrapolation, generating suitability maps that support targeted surveillance, identify areas at risk of mosquito establishment, and inform proactive vector control. The framework also maintains predictive performance under novel climatic conditions, highlighting its potential for projecting mosquito distributions under future climate change scenarios.