Global spatial risk pattern of highly pathogenic avian influenza H5N1 virus in wild birds: A knowledge-fusion based approach.

Highly pathogenic avian influenza (HPAI) H5N1 viruses have continuously circulated throughout much of the world since 2003, resulting in huge economic losses and major public health problems. Wild birds have played an important role in the spread of H5N1 HPAI. To understand its spatial distribution, H5N1 HPAI have been studied by many disciplines from different perspectives, but only one kind of disciplinary knowledge was involved, which has provided limited progress in understanding. Combining risk information from different disciplines based on knowledge fusion can provide more accurate and detailed information. In this study, local k function, phylogenetic tree analysis, and logistic spatial autoregressive models were used to explore the global spatial pattern of H5N1 HPAI based on outbreak data in wild birds, genetic sequences, and risk factors, respectively. On this basis, Dempster-Shafer (D-S) evidence theory was further applied to study the spatial distribution of H5N1 HPAI. We found D-S evidence theory was more robust and reliable than the other three methods, providing technical and methodological support for application to the research of other diseases. The shortest distance to wild bird migration routes, roads and railways, elevation, the normalized difference vegetation index (NDVI), land use and land cover (LULC) and infant mortality rates (IMR) were significantly associated with the occurrence of H5N1 HPAI. The high-risk areas were mainly located in Northern and Central Europe, the eastern Mediterranean, and East and Southeast Asia. High-risk clusters were closely related to the social, economic and ecological environment of the region. Locations where the potential transmission risk remains high should be prioritized for control efforts.