A STATISTICAL ANALYSIS OF THE COVID-19 OUTBREAK DATA
The exponential model is a commonly used epidemic model for the analysis of initial outbreak data due to an infectious disease. But there have been questions about its validity in practice. This article examines this issue through statistical analysis on 22 countries’ initial COVID-19 outbreak data provided by the World Health Organization. For each of 22 countries, a general regression analysis is conducted for the cumulative confirmed cases. Our regression function is a 3-5 piecewise fitted functions which are obtained via regression analysis, data transformation and careful detection of change-points. The detection of change-points is conducted by the combination of residual analysis and r2 values, and we introduce the concept of overall R2 value to assess the goodness of fit for the fitted curves. Among the 22 countries considered in our analysis, 77.3% and 22.7% of these countries had fitted exponential curve and logistic curve, respectively, during the initial period. The average length of the fitted exponential curves is 31 days, and the average length of the fitted logistic curves is 39 days. Regionally, we have similar results. In conclusion, our data analysis in this article suggests that the usual epidemic exponential model assumption does not always hold in practice.
change-point, exponential model, logistic model, pandemic data, regression analysis, residual analysis.
Received: February 5, 2023; Accepted: March 15, 2023; Published: March 20, 2023
How to cite this article: Jian-Jian Ren, Yiming Lyu, Chen Qian, Yuyin Shi and Charles Zhao, A statistical analysis of the COVID-19 outbreak data, JP Journal of Biostatistics 23(1) (2023), 77-94. http://dx.doi.org/10.17654/0973514323005
This Open Access Article is Licensed under Creative Commons Attribution 4.0 International License
References:
[1] Alvin C. Rencher, Linear Models in Statistics, Wiley, 1st ed., 1999.[2] J. Neter, W. Wasserman and M. H. Kutner, Applied linear statistical models, Regression, Analysis of Variance and Experimental Designer’s, Richard D. Irwin, INC, 2nd ed., 1985.[3] Franklin A. Graybill, Theory and Application of the Linear Model, Wadsworth & Brooks/Cole, 1976.[4] Odile Pons, Estimation and Tests in Change-point Models, World Scientific, 2018.