Premium
Modelling SARS data using threshold geometric process
Author(s) -
Chan Jennifer S. K.,
Yu Philip L. H.,
Lam Yeh,
Ho Alvin P. K.
Publication year - 2006
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2376
Subject(s) - monotone polygon , covid-19 , stage (stratigraphy) , limit (mathematics) , mathematics , statistics , process (computing) , time point , computer science , disease , medicine , mathematical analysis , physics , infectious disease (medical specialty) , geometry , biology , paleontology , pathology , operating system , acoustics
Abstract During the outbreak of an epidemic disease, for example, the severe acute respiratory syndrome (SARS), the number of daily infected cases often exhibit multiple trends: monotone increasing during the growing stage, stationary during the stabilized stage and then decreasing during the declining stage. Lam first proposed modelling a monotone trend by a geometric process (GP) { X i , i =1,2,…} directly such that { a i −1 X i , i =1,2,…} forms a renewal process for some ratio a >0 which measures the direction and strength of the trend. Parameters can be conveniently estimated using the LSE methods. Previous GP models limit to data with only a single trend. For data with multiple trends, we propose a moving window technique to locate the turning point(s). The threshold GP model is fitted to the SARS data from four regions in 2003. Copyright © 2005 John Wiley & Sons, Ltd.