Detecting Dengue in Flight: Leveraging Machine Learning to Analyze Mosquito Flight Patterns for Infection Detection.

Abstract

With the growing global threat of mosquito-borne diseases, there is an urgent need for faster, automated methods to assess disease load in mosquitoes and predict outbreaks. Current surveillance relies on manual mosquito traps and labor-intensive lab tests like polymerase chain reaction (PCR) and enzyme-linked immunosorbent assay (ELISA), which are time-consuming and resource-intensive. In this study, machine learning algorithms are applied to detect dengue-infected mosquitoes based on their 3D flight patterns. Using a convolutional neural network (CNN) and cubic spline interpolation, mosquito flight trajectories are tracked, followed by classification with models including CNN, eXtreme Gradient Boosting (XGBoost), Adaptive Boosting (AdaBoost), Random Forest, Decision Tree, Naive Bayes, Logistic Regression, Multi-Layer Perceptron (MLP), and a hybrid CNN + XGBoost model. The 5-fold cross-validation results showed that XGBoost achieved the highest mean accuracy (81.43%), while Random Forest has shown the best mean F1 Score (82.80%). Some validation folds demonstrated outstanding performance, with AdaBoost reaching 95.85% accuracy and Random Forest achieving 97.77% recall in Fold 1. The study also analyzed the impact of flight sequence size on models' performance, observing that longer sequences provided more accurate predictions. This approach offers a faster and more efficient method for assessing infection status, supporting real-time vector monitoring, and improving early disease outbreak detection.