Epidemiological model including spatial connection features improves prediction of the spread of pine wilt disease

Abstract

In recent decades, the tree killing pine wilt disease (PWD) has been spreading globally and has caused substantial damage, particularly in China, resulting in annual losses of nearly $15 billion. This disease is caused by the pine wood nematode (Bursaphelenchus xylophilus) and is transmitted by its vector insect (Monochamus spp.). However, there remains limited understanding of the factors influencing PWD transmission, especially anthropogenic factors. For the first time, we employed a dataset of propagation factors that based on the transportation of pine wood transmission and used an enhanced infectious disease dynamics method to quantitatively predict the likelihood of future PWD infections across the entire distribution range of pine trees in China. The validation results show that the risk of infection in potential areas of PWD will further increase in China, with the average risk surpassing 30% in 2023 and reaching 50% by 2025 in the study areas. Meanwhile, we observe that human factors, such as transportation of timber, are becoming the main cause of the rapid spread of PWD and require extra vigilance. By incorporating spatial connections between samples as features for PWD prediction, the Area Under the Curve (AUC) score of the model reached 0.90. This signifies a highly accurate prediction of the pivotal areas for PWD dissemination, providing valuable guidance for the precise development of prevention and control measures, and presenting innovative approaches for analogous studies.