SENTINEL-Dengue ASEAN-11: An AI-Powered Climate–Geospatial Intelligence System for Dengue Early Warning and Surveillance Prioritization Across Southeast Asia

Abstract

Dengue remains a major climate-sensitive vector-borne disease in tropical and subtropical regions, particularly across Southeast Asia where rainfall variability, temperature suitability, vegetation dynamics, urbanization, surface water, and population exposure interact to shape transmission risk. Conventional dengue surveillance is often reactive and administratively aggregated, limiting its usefulness for early warning, targeted vector control, and regional preparedness. This study develops SENTINEL-Dengue ASEAN-11, an artificial intelligence and climate–geospatial intelligence framework for dengue early warning and surveillance prioritization across Brunei Darussalam, Cambodia, Indonesia, Lao PDR, Malaysia, Myanmar, the Philippines, Singapore, Thailand, Timor-Leste, and Viet Nam. The framework integrates publicly available dengue case-count data, climate reanalysis, remote-sensing vegetation indices, gridded population exposure, surface-water indicators, administrative boundaries, explainable artificial intelligence, spatial block cross-validation, temporal validation, and uncertainty mapping. A harmonized Admin-1 month geospatial panel dataset is proposed for 2010–2023. Dengue alert outcomes are defined using within-area historical baselines to reduce misleading country-level comparison. Statistical models, Bayesian spatiotemporal smoothing, Random Forest, XGBoost, and explainable AI are used to estimate dengue alert probability and identify influential climate–environmental predictors. The final output is an ASEAN-11 surveillance-priority framework that distinguishes high-risk/high-confidence, high-risk/high-uncertainty, moderate-risk, low-risk, and data-insufficient areas. SENTINEL-Dengue ASEAN-11 advances dengue geoinformatics beyond retrospective hotspot mapping toward interpretable, scalable, and policy-relevant epidemic intelligence for climate-sensitive public-health preparedness.