Distribution dynamics and urbanization-related factors of Hantaan and Seoul virus infections in China between 2001 and 2020: A machine learning modelling analysis

Abstract

Objectives: The epidemical and clinical features of distinct hantavirus infections exhibit heterogeneity. However, the evolving epidemics and distinct determines of the two hantavirus infections remain uncertain. Methods: Data on hemorrhagic fever with renal syndrome (HFRS) cases and genotyping were collected from multiple sources to explore the distribution dynamics of different endemic categories. Four modelling algorithms were used to examine the relationship between infected hantavirus genotypes in HFRS patients, as well as assess the impacts of urbanization-related factors on HFRS incidence. Results: The number of cities dominated by Hantaan (HTNV) and Seoul (SEOV) viruses was projected to decrease between two phases, while the mixed endemic cities increased. Patients with SEOV infection predominantly presented gastrointestinal symptoms. The modeling analysis revealed that built-up land and real GDP demonstrated the highest contribution to HTNV and SEOV infections, respectively. The impact of nightlight index and park green land was more pronounced in HTNV-dominant cities, while cropland, impervious surface, and floor space of commercialized buildings sold contributed more to HFRS incidence in SEOV-dominant cities. Conclusions: Our findings fill a gap for the three endemic categories of HFRS, which may guide the development of targeted prevention and control measures under the conditions of urbanization development.

Keywords

• HFRS
• XGBoost
• SHAP
• Urbanization
• Risk factor