Innovative Ensemble Approaches for Assessing Critical Factors for European Tick Abundance

This research uses complex machine learning methods to explore the driving factors behind increasing Ixodes ricinus populations in Europe. This tick species plays a key role in transmission of tick-borne diseases, which present a threat to both human and animal health.

Tick prevalence rates were examined alongside climatic features (temperature and rainfall) and habitat features (land use category and vegetation levels). The data were analysed using Ensemble modelling approaches (XGBoost, LightGBM, CatBoost, Voting Regressor, Bagging Regressor, Stacking Regressor and AdaBoost Regressor) which were selected for their ability to embrace this complex and multifactorial dataset. The predictive performance of these models was carefully evaluated using Root Mean Square Error (RMSE) and R-squared (R2) (both in the presence and absence of outliers).

Our results revealed that land use was the dominant factor for predicting tick occurrences. Stacking Regressor showed the highest performance, achieving an RMSE of 2.34 and an R² value of 0.99. After removing outliers, the models showed better results with a decrease in average RMSE from 13.14 to 7.92 and R² increase from 0.21 to 0.81 suggesting that outlier management impacts upon model performance.

Our machine learning algorithm accurately predicts tick abundance and provides a key insight into which areas are most likely to be at risk, guiding future location-specific intervention to combat these risks, which will ultimately act as a powerful tool in the fight against tick-borne diseases.

Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.