Global Spatiotemporal Dynamics of African Swine Fever: An Integrated Multi-Scale Spatial and Time-Series Analysis

Area of Science:

• Veterinary epidemiology
• Disease ecology
• Predictive modeling

Background:

• African swine fever (ASF) presents a significant and growing global threat to swine production.
• Understanding its spatiotemporal dynamics is crucial for effective control.

Purpose of the Study:

• To characterize the global spatiotemporal dynamics of ASF from 1996 to 2025.
• To compare the predictive performance of various forecasting models for ASF outbreaks.

Main Methods:

• Integrated framework combining multi-distance spatial analysis and time series forecasting.
• Utilized a dataset of 57,253 ASF outbreak records.
• Evaluated machine learning, statistical, and deep learning models (Random Forest, SARIMA, Prophet, XGBoost, GRU).

Main Results:

• Wild boar were responsible for ~70% of outbreaks, primarily in Eastern Europe; domestic pig outbreaks concentrated in Southeast Asia.
• ASF spread shifted from a North-South to an East-West axis across Eurasia (2017-2020).
• Bimodal seasonal patterns observed in the Northern Hemisphere (Jan-Mar, Jul-Aug).
• Random Forest model demonstrated superior predictive accuracy over other tested models.

Conclusions:

• Wild boar play a critical role in sustaining global ASF transmission.
• Integrated surveillance at wildlife-livestock interfaces is essential.
• Machine learning approaches, particularly Random Forest, can enhance ASF early warning systems and control strategies.