Abstract:
Climate change is fundamentally reshaping the epidemiological landscape, with approximately 75% of emerging infectious diseases originating from zoonotic events. In Southeast As a region characterized by high population density, intensive livestock production, and extreme climatic variability. The convergence of agricultural expansion, deforestation, and climate anomalies such as the El Nino Southern Oscillation has created a high-risk environment for disease emergence. Traditional reactive surveillance systems have proven inadequate, as fragmented data across human, animal, and environmental sectors delays outbreak detection and response. This presentation proposes a predictive Early Warning System (EWS) grounded in the One Health paradigm, designed to anticipate climate sensitive zoonotic outbreaks in livestock systems across the Asia-Pacific region.
The proposed EWS architecture integrates four core data layers:
(i) Earth observation and climatic data (precipitation, land surface temperature, soil moisture) from platforms such as Copernicus and NASA Earth Observation.
(ii) animal mobility and wildlife migration tracking.
(iii) historical epidemiologicalrecords and seroprevalence data from sentinel populations.
(iv) land use change indicators, including deforestation and agricultural frontier expansion.
These inputs are processed through an Extract Transform Load pipeline and analyzed using an ensemble of machine learning models Random Forest, XGBoost, and Long Short Term Memory networks to compute a dynamic Zoonotic Risk Index (ZRI). The system is calibrated using historical outbreak data from Thailand, Vietnam, and Malaysia (2015–2025), and validated against known epidemic events. A Decision Support System (DSS) interface translates probabilistic outputs into actionable alerts for veterinary, agricultural, and public health authorities. Retrospective validation demonstrates that the EWS achieves 85% sensitivity in detecting historical outbreaks with a lead time of 48–72 hours, and an area under the ROC curve of 0.92. The analysis revealed that anomalous precipitation patterns combined with deforestation in water catchment areas increased vector borne disease risk in swine farms by 40%. The DSS successfully generated region specific recommendations, enabling targeted quarantine measures and optimized sampling protocols. These findings align with emerging regional initiatives, such as Singapore’s AI-powered epidemiological intelligence platform and ASEAN wide surveillance frameworks, which emphasize predictive analytics and cross-border data sharing. Despite its demonstrated efficacy, several adoption barriers remain, including data quality gaps in rural areas, interoperability challenges across national surveillance systems, and limited digital literacy among field level stakeholders. To address these, the presentation proposes a roadmap toward 2030 that includes the integration of digital twins for scenario simulation, the use of generative AI to translate complex alerts into local-language recommendations, and the establishment of regional data-sharing agreements under the One Health framework. This work concludes that predictive analytics are not a substitute for conventional biosecurity measures but a strategic complement essential for building climate-resilient and food-secure agricultural systems. As evidenced by FAO’s Rift Valley Fever Early Warning Decision Support Tool and similar initiatives, integrating Earth observation, mobility data, and machine learning into epidemiological surveillance significantly enhances preparedness and response capacity. Investment in data infrastructure and cross sector collaboration is therefore critical to safeguarding both livestock productivity and global health security in an era of accelerating climate change.