Artificial Intelligence and Machine Learning for Mosquito Population Forecasting: A Systematic Review and Meta-analysis of Predictive Performance and Methodological Quality
1
Faculty of Medicine, University Vita-Salute San Raffaele, Milan, Italy
2
PhD National Program in One Health Approaches to Infectious Diseases and Life Science Research, Department of Public Health, Experimental and Forensic Medicine, University of Pavia, Pavia, Italy
3
Department of Cardiac Thoracic Vascular Sciences and Public Health, University of Padua, Padova, Italy
Popul. Med. 2026;8(Supplement Supplement 1):A3825
ABSTRACT
INTRODUCTION:
Mosquitoes of the genera Aedes, Culex, and Anopheles transmit major infectious diseases and represent a growing global health threat. Forecasting their population dynamics is crucial for early warning and vector control. Although artificial intelligence (AI) and machine learning (ML) models are increasingly used for this purpose, evidence is scattered and no prior quantitative synthesis has evaluated their predictive performance and methodological quality. This study addresses that gap through a systematic review and meta-analysis.
METHODS:
Following PRISMA 2020 guidelines and a PROSPERO-registered protocol, we searched PubMed, Embase, and Scopus through August 2025 for original studies using AI or ML to predict mosquito distribution, abundance, or dynamics. We extracted modelling characteristics and performance metrics and assessed risk of bias with PROBAST. We conducted a meta-analysis pooling AUC, accuracy, sensitivity, specificity, PPV, and NPV using fixed- and random-effects models.
RESULTS:
Seventy-six studies met inclusion criteria, mainly from Asia, North America, and Europe, most frequently targeting Aedes aegypti and Aedes albopictus. Environmental variables were the dominant predictors (n=71). Random Forest (n=27), MaxEnt (n=10), and XGBoost (n=5) were the most commonly used algorithms, with growing adoption of deep learning for image and time-series data. Predictive performance was generally high, with AUC often above 0.90. Pooled estimates indicate that ML models outperform or match traditional statistical approaches (0.79 AUC). However, methodological limitations were common: about 55% of studies had high risk of bias, and only 12 of 76 reported external validation. Most models remained at a research-only stage.
CONCLUSIONS:
This work fills a critical evidence gap by delivering the first meta-analytic synthesis of AI/ML mosquito prediction models and their methodological robustness. While ML approaches show superior predictive performance, weak validation practices and frequent bias limit translational reliability. Standardized validation and reporting frameworks are necessary to support operational deployment.
Mosquitoes of the genera Aedes, Culex, and Anopheles transmit major infectious diseases and represent a growing global health threat. Forecasting their population dynamics is crucial for early warning and vector control. Although artificial intelligence (AI) and machine learning (ML) models are increasingly used for this purpose, evidence is scattered and no prior quantitative synthesis has evaluated their predictive performance and methodological quality. This study addresses that gap through a systematic review and meta-analysis.
METHODS:
Following PRISMA 2020 guidelines and a PROSPERO-registered protocol, we searched PubMed, Embase, and Scopus through August 2025 for original studies using AI or ML to predict mosquito distribution, abundance, or dynamics. We extracted modelling characteristics and performance metrics and assessed risk of bias with PROBAST. We conducted a meta-analysis pooling AUC, accuracy, sensitivity, specificity, PPV, and NPV using fixed- and random-effects models.
RESULTS:
Seventy-six studies met inclusion criteria, mainly from Asia, North America, and Europe, most frequently targeting Aedes aegypti and Aedes albopictus. Environmental variables were the dominant predictors (n=71). Random Forest (n=27), MaxEnt (n=10), and XGBoost (n=5) were the most commonly used algorithms, with growing adoption of deep learning for image and time-series data. Predictive performance was generally high, with AUC often above 0.90. Pooled estimates indicate that ML models outperform or match traditional statistical approaches (0.79 AUC). However, methodological limitations were common: about 55% of studies had high risk of bias, and only 12 of 76 reported external validation. Most models remained at a research-only stage.
CONCLUSIONS:
This work fills a critical evidence gap by delivering the first meta-analytic synthesis of AI/ML mosquito prediction models and their methodological robustness. While ML approaches show superior predictive performance, weak validation practices and frequent bias limit translational reliability. Standardized validation and reporting frameworks are necessary to support operational deployment.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
