Publication: A mixture model application in disease mapping of malaria
Issued Date
2004-03-01
Resource Type
ISSN
01251562
Other identifier(s)
2-s2.0-3042747062
Rights
Mahidol University
Rights Holder(s)
SCOPUS
Bibliographic Citation
Southeast Asian Journal of Tropical Medicine and Public Health. Vol.35, No.1 (2004), 38-47
Suggested Citation
Sasivimol Rattanasiri, Dankmar Böhning, Piangchan Rojanavipart, Suthi Athipanyakom A mixture model application in disease mapping of malaria. Southeast Asian Journal of Tropical Medicine and Public Health. Vol.35, No.1 (2004), 38-47. Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/21719
Research Projects
Organizational Units
Authors
Journal Issue
Thesis
Title
A mixture model application in disease mapping of malaria
Other Contributor(s)
Abstract
Disease mapping, a method for displaying the geographical distribution of disease occurrence, has received attention for more than 2 decades. Because traditional approaches to disease mapping have some deficiencies and disadvantages in presenting the geographical distribution of disease, the mixture model-as an alternative approach-overcomes some of these deficiencies and provides a clearer picture of the spatial risk structure. The purpose of this study was twofold: (1) to investigate the geographical distribution of malaria in Thailand during 1995, 1996, and 1997 by applying the mixture model to disease mapping, and (2) to investigate the dynamic nature of malaria in Thailand during the 3-year time frame by applying the space-time mixture model. Non-parametric maximum likelihood estimation was employed to estimate the parameters of both the mixture model and the space-time mixture model. Applying Bayes' theorem, the 76 provinces of Thailand were classified into component risk levels by the rate of malaria for each province. Malaria intensively occurred in 4 provinces on the Thai-Myanmar border and in 2 provinces on the Thai-Cambodian border. Of the 76 provinces studied, 10 showed an increasing trend over the 3-year period. A comparison of the map based on the mixture model with the map based on the traditional percentiles method indicates that the non-parametric mixture model removes random variability from the map and provides a clearer picture of the spatial risk structure. The advantage of the mixture model approach to disease mapping is the graphical visual presentation of the prevalence of disease. The space-time mixture model more adequately investigates the dynamic nature of disease than does the mixture model.