Extreme Value Modeling for the Rate of Covid-19 in the Northeast of Thailand
Keywords:
covid 19, covidgeneralized extreme value distribution, stationary process, non-stationary processAbstract
The objective of this study was to develop an extreme value model to analyze and predict the COVID-19 morbidity rate in northeastern Thailand. Weekly maximum case counts from 20 provinces were used to calculate the weekly maximum morbidity rate (cases per 100,000 population). The analysis employed the Generalized Extreme Value (GEV) distribution, with parameter estimation performed using the Maximum Likelihood Estimation (MLE) method under both stationary and non-stationary processes across eight model structures. Model selection was based on deviance statistics and negative log-likelihood values, with goodness-of-fit assessed using probability and quantile plots. In addition, return levels were calculated to estimate recurrence intervals for extreme morbidity rates. The results indicated that the non-stationary model, in which the location parameter depends on covariates and while the shape parameter is held constant, provided the best fit across all 20 provinces. Udon Thani exhibited the highest risk, with the return level reaching approximately 60 cases per 100,000 population for a rare outbreak event occurring once every 40 weeks (p = 0.05), while more frequent outbreaks corresponded to return levels of 13–42 cases per 100,000 population, suggesting the need for focused monitoring and public health interventions in this area.
References
Wikipedia contributors. COVID-19 pandemic in Thailand [Internet]. Wikipedia; 2022 [cited 2025 Sep 10]. Available from:https://th.wikipedia.org/wiki/
Coles S. An Introduction to Statistical Modeling of Extreme Values. London: Springer; 2004.
Villeneuve PJ, Weichenthal S, Crouse D, Miller AB, To T, Martin RV, et al. Long-term exposure to fine particulate matter air pollution and mortality among Canadian women. Epidemiology. 2015;26(4):536-45.
Reinhart A. An application of extreme value theory to the monitoring of influenza epidemics. arXiv [Preprint]. 2019 Oct 23. arXiv:1910.10788. Available from: https://arxiv.org/abs/1910.10788
Wang X, Gao Z, Zhang J. Modeling the impact of temperature extremes on mortality using bivariate peaks-over-threshold and ARIMA approaches. BMC Public Health. 2024;24:1586.
Alves I. Mortality and morbidity peaks modeling: An extreme value theory approach [master’s thesis]. Porto (PT): University of Porto; 2016.
Gampe J. Human mortality beyond age 110. In: Maier H, Jeune B, Vaupel JW, editors. Exceptional Longevity: From Prehistory to the Present. Berlin: Springer; 2006. p. 219-30.
Rootzén H, Zholud D. Human life is unlimited – but short. Extremes. 2017; 20(4):713-28.
Gilleland E, Katz RW. extRemes 2.0: An extreme value analysis package in R. J Stat Softw. 2016;72(8):1-39. doi:10.18637/ jss.v072.i08.
Ministry of Public Health (TH), Department of Disease Control. COVID-19 dashboard [Internet]. Bangkok: MOPH; 2022 [cited 2022 Mar 20]. Available from: https://covid19.ddc.moph.go.th/
Bauch CT, Lloyd-Smith JO, Coffee MP, Galvani AP. Dynamically modeling SARS and other newly emerging respiratory illnesses: past, present, and future. Epidemiology. 2005;16(6):791-801.
Chowell G, Abdirizak F, Lee S, Lee J, Jung E, Nishiura H, et al. Transmission characteristics of MERS and SARS in the healthcare setting: a comparative study. BMC Med. 2015;13:210.
Nishiura H. Early efforts in modeling the incubation period of influenza A (H1N1) and its application for surveillance. Euro Surveill. 2009;14(13):19193
Downloads
Published
How to Cite
Issue
Section
License
This article was published in the Journal of regional healh promotion centre 7 khonkaen. It is considered an academic work or research. The results of the analysis and recommendations are subjective opinions. It is not the opinion of the Journal of regional healh promotion centre 7 khonkaen or the editorial office in any way. Authors are responsible for their own articles.
