tetano
Editor, Senior Moderator
J Vet Res
. 2021 Jul 5;65(3):253-264.
doi: 10.2478/jvetres-2021-0034. eCollection 2021 Sep.
Regional Distribution of Non-human H7N9 Avian Influenza Virus Detections in China and Construction of a Predictive Model
Zeying Huang[SUP] 1 [/SUP], Haijun Li[SUP] 2 [/SUP], Beixun Huang[SUP] 1 [/SUP]
Affiliations
Abstract
Introduction: H7N9 avian influenza has broken out in Chinese poultry 10 times since 2013 and impacted the industry severely. Although the epidemic is currently under control, there is still a latent threat.
Material and methods: Epidemiological surveillance data for non-human H7N9 avian influenza from April 2013 to April 2020 were used to analyse the regional distribution and spatial correlations of positivity rates in different months and years and before and after comprehensive immunisation. In addition, positivity rate monitoring data were disaggregated into a low-frequency and a high-frequency trend sequence by wavelet packet decomposition (WPD). The particle swarm optimisation algorithm was adopted to optimise the least squares support-vector machine (LS-SVM) model parameters to predict the low-frequency trend sequence, and the autoregressive integrated moving average (ARIMA) model was used to predict the high-frequency one. Ultimately, an LS-SVM-ARIMA combined model based on WPD was constructed.
Results: The virus positivity rate was the highest in late spring and early summer, and overall it fell significantly after comprehensive immunisation. Except for the year 2015 and the single month of December from 2013 to 2020, there was no significant spatiotemporal clustering in cumulative non-human H7N9 avian influenza virus detections. Compared with the ARIMA and LS-SVM models, the LS-SVM-ARIMA combined model based on WPD had the highest prediction accuracy. The mean absolute and root mean square errors were 2.4% and 2.0%, respectively.
Conclusion: Low error measures prove the validity of this new prediction method and the combined model could be used for inference of future H7N9 avian influenza virus cases. Live poultry markets should be closed in late spring and early summer, and comprehensive H7N9 immunisation continued.
Keywords: LS-SVM-ARIMA model; PSO algorithm; aetiological virus detection monitoring; epidemiological survey; wavelet packet decomposition.
. 2021 Jul 5;65(3):253-264.
doi: 10.2478/jvetres-2021-0034. eCollection 2021 Sep.
Regional Distribution of Non-human H7N9 Avian Influenza Virus Detections in China and Construction of a Predictive Model
Zeying Huang[SUP] 1 [/SUP], Haijun Li[SUP] 2 [/SUP], Beixun Huang[SUP] 1 [/SUP]
Affiliations
- PMID: 34917836
- PMCID: PMC8643092
- DOI: 10.2478/jvetres-2021-0034
Abstract
Introduction: H7N9 avian influenza has broken out in Chinese poultry 10 times since 2013 and impacted the industry severely. Although the epidemic is currently under control, there is still a latent threat.
Material and methods: Epidemiological surveillance data for non-human H7N9 avian influenza from April 2013 to April 2020 were used to analyse the regional distribution and spatial correlations of positivity rates in different months and years and before and after comprehensive immunisation. In addition, positivity rate monitoring data were disaggregated into a low-frequency and a high-frequency trend sequence by wavelet packet decomposition (WPD). The particle swarm optimisation algorithm was adopted to optimise the least squares support-vector machine (LS-SVM) model parameters to predict the low-frequency trend sequence, and the autoregressive integrated moving average (ARIMA) model was used to predict the high-frequency one. Ultimately, an LS-SVM-ARIMA combined model based on WPD was constructed.
Results: The virus positivity rate was the highest in late spring and early summer, and overall it fell significantly after comprehensive immunisation. Except for the year 2015 and the single month of December from 2013 to 2020, there was no significant spatiotemporal clustering in cumulative non-human H7N9 avian influenza virus detections. Compared with the ARIMA and LS-SVM models, the LS-SVM-ARIMA combined model based on WPD had the highest prediction accuracy. The mean absolute and root mean square errors were 2.4% and 2.0%, respectively.
Conclusion: Low error measures prove the validity of this new prediction method and the combined model could be used for inference of future H7N9 avian influenza virus cases. Live poultry markets should be closed in late spring and early summer, and comprehensive H7N9 immunisation continued.
Keywords: LS-SVM-ARIMA model; PSO algorithm; aetiological virus detection monitoring; epidemiological survey; wavelet packet decomposition.