Mapping Mechanistic Modeling of Highly Pathogenic Avian Influenza Across Migratory Flyways: A Systematic Review

Zubair Ahmad , Hao Wang, Simon Plakolb, Robin N. Thompson, Nils Chr. Stenseth

Highly pathogenic avian influenza (HPAI) viruses, particularly H5N1 clade 2.3.4.4b, continue to spread globally via wild migratory birds along defined flyway corridors. Mechanistic models are essential tools for understanding HPAI transmission dynamics in flyway systems. Yet the geographic distribution of such modeling efforts across migratory flyways remains unknown. To inform the global response to HPAI, we conducted a PRISMA-compliant systematic review of published mechanistic and mathematical models of HPAI that explicitly incorporate migratory wild bird populations, organizing findings within a flyway-stratified framework. A search of PubMed, Scopus, Web of Science, and Embase yielded 582 records, of which 30 met inclusion criteria: 20 HPAI-specific models (H5N1, H5N8, H5N6, multiple HPAI subtypes) and 10 general avian influenza virus (AIV) models or pre-2016 H7N9 studies retained for their transmission dynamics relevance. Included studies spanned 2010–2025, with a marked surge in 2025 following worldwide transmission of HPAI in the early 2020s. The East Asian–Australasian Flyway (EAAF) dominated modeling effort (35.6% of studies), followed by the East Atlantic (15%), Central Asian (12.2%), Mediterranean–Black Sea (11.7%), and Americas flyways (10%). American (10%), European (27%) and African flyways (2.2%) were underrepresented, despite carrying millions of migratory waterfowl annually and experiencing unprecedented HPAI activity since 2020. Compartmental ordinary differential equation (ODE) models were the most common framework (33.3%) used across the included studies, followed by agent-based and spatial models (20% each). H5N1dominated subtype coverage (53.3%). We identified several recurring methodological simplifications: models typically assumed homogeneous mixing within populations, rarely linked wild bird reservoirs to domestic livestock such as poultry and human spillover in full transmission chains, and underutilized available empirical tracking data. Our findings reveal a systematic geographic asymmetry in HPAI mechanistic modeling that misaligns with current global outbreak distributions, with direct implications for early warning system development and international surveillance coordination.