Development of a causal machine learning model for the diagnosis of African swine fever

Abstract

This study investigates the causal relationship between African swine fever (ASF) viral load and disease severity in domestic and wild pigs using machine learning models. A causality model with linear regression and random forest regressor was developed to analyse ASF transmission dynamics and symptom severity. The linear regression model achieved an R² value of 83.68% with an MAE of 1.27, while the random forest model achieved an R² value of 58.10% with an MAE of 1.52, confirming strong predictive performance. The results highlight the effectiveness of biosecurity, surveillance and culling measures in containing ASF and em phasize evidence-based policy making for disease control. This study provides actionable in sights for veterinarians, farmers and policy makers, contributing to ASF risk management and prevention strategies. Future research should integrate AI-driven real-time surveillance and genetic analysis to improve ASF outbreak prediction and global containment measures.