tetano
Editor, Senior Moderator
J Infect Public Health
. 2024 Apr 30;17(6):1086-1094.
doi: 10.1016/j.jiph.2024.04.024. Online ahead of print. Global pattern and determinant for interaction of seasonal influenza viruses
Yilin Chen[SUP] 1 [/SUP], Feng Tang[SUP] 2 [/SUP], Zicheng Cao[SUP] 3 [/SUP], Jinfeng Zeng[SUP] 1 [/SUP], Zekai Qiu[SUP] 1 [/SUP], Chi Zhang[SUP] 1 [/SUP], Haoyu Long[SUP] 1 [/SUP], Peiwen Cheng[SUP] 1 [/SUP], Qianru Sun[SUP] 1 [/SUP], Wenjie Han[SUP] 1 [/SUP], Kang Tang[SUP] 1 [/SUP], Jing Tang[SUP] 1 [/SUP], Yang Zhao[SUP] 4 [/SUP], Dechao Tian[SUP] 4 [/SUP], Xiangjun Du[SUP] 5 [/SUP]
Affiliations
Background: The prevalence of different types/subtypes varies across seasons and countries for seasonal influenza viruses, indicating underlying interactions between types/subtypes. The global interaction patterns and determinants for seasonal influenza types/subtypes need to be explored.
Methods: Influenza epidemiological surveillance data, as well as multidimensional data that include population-related, environment-related, and virus-related factors from 55 countries worldwide were used to explore type/subtype interactions based on Spearman correlation coefficient. The machine learning method Extreme Gradient Boosting (XGBoost) and interpretable framework SHapley Additive exPlanation (SHAP) were utilized to quantify contributing factors and their effects on interactions among influenza types/subtypes. Additionally, causal relationships between types/subtypes were also explored based on Convergent Cross-mapping (CCM).
Results: A consistent globally negative correlation exists between influenza A/H3N2 and A/H1N1. Meanwhile, interactions between influenza A (A/H3N2, A/H1N1) and B show significant differences across countries, primarily influenced by population-related factors. Influenza A has a stronger driving force than influenza B, and A/H3N2 has a stronger driving force than A/H1N1.
Conclusion: The research elucidated the globally complex and heterogeneous interaction patterns among influenza type/subtypes, identifying key factors shaping their interactions. This sheds light on better seasonal influenza prediction and model construction, informing targeted prevention strategies and ultimately reducing the global burden of seasonal influenza.
Keywords: Causality; Correlation; Interaction; Machine learning; Seasonal influenza types/subtypes.
. 2024 Apr 30;17(6):1086-1094.
doi: 10.1016/j.jiph.2024.04.024. Online ahead of print. Global pattern and determinant for interaction of seasonal influenza viruses
Yilin Chen[SUP] 1 [/SUP], Feng Tang[SUP] 2 [/SUP], Zicheng Cao[SUP] 3 [/SUP], Jinfeng Zeng[SUP] 1 [/SUP], Zekai Qiu[SUP] 1 [/SUP], Chi Zhang[SUP] 1 [/SUP], Haoyu Long[SUP] 1 [/SUP], Peiwen Cheng[SUP] 1 [/SUP], Qianru Sun[SUP] 1 [/SUP], Wenjie Han[SUP] 1 [/SUP], Kang Tang[SUP] 1 [/SUP], Jing Tang[SUP] 1 [/SUP], Yang Zhao[SUP] 4 [/SUP], Dechao Tian[SUP] 4 [/SUP], Xiangjun Du[SUP] 5 [/SUP]
Affiliations
- PMID: 38705061
- DOI: 10.1016/j.jiph.2024.04.024
Background: The prevalence of different types/subtypes varies across seasons and countries for seasonal influenza viruses, indicating underlying interactions between types/subtypes. The global interaction patterns and determinants for seasonal influenza types/subtypes need to be explored.
Methods: Influenza epidemiological surveillance data, as well as multidimensional data that include population-related, environment-related, and virus-related factors from 55 countries worldwide were used to explore type/subtype interactions based on Spearman correlation coefficient. The machine learning method Extreme Gradient Boosting (XGBoost) and interpretable framework SHapley Additive exPlanation (SHAP) were utilized to quantify contributing factors and their effects on interactions among influenza types/subtypes. Additionally, causal relationships between types/subtypes were also explored based on Convergent Cross-mapping (CCM).
Results: A consistent globally negative correlation exists between influenza A/H3N2 and A/H1N1. Meanwhile, interactions between influenza A (A/H3N2, A/H1N1) and B show significant differences across countries, primarily influenced by population-related factors. Influenza A has a stronger driving force than influenza B, and A/H3N2 has a stronger driving force than A/H1N1.
Conclusion: The research elucidated the globally complex and heterogeneous interaction patterns among influenza type/subtypes, identifying key factors shaping their interactions. This sheds light on better seasonal influenza prediction and model construction, informing targeted prevention strategies and ultimately reducing the global burden of seasonal influenza.
Keywords: Causality; Correlation; Interaction; Machine learning; Seasonal influenza types/subtypes.