|Return to page on Research in Tuberculosis Control|
Bayesian Mixture Analysis for Tuberculin Induration Data
The analysis of tuberculin skin test survey data is complicated due to limitations of the operating characteristics of the tuberculin skin test. Mixture analysis allows analyzing data arising from different sub-groups (such as reactions attributable to Mycobacterium tuberculosis infection and reactions attributable to sensitization resulting from infection with environmental mycobacteria).
Upon request of The Union, Dr Beat E Neuenschwander has developed, with the financial support of The Union, a series of programs that can be utilized to examine tuberculin skin test survey data with mixture analysis.
|Before you commence any analysis, you should carefully read the documentation (a PDF file which you obtain by clicking on the icon to the left: 593KB).|
|Obtain the scripts and the sample dataset by clicking on the icon to the left. Note that this is a self-extracting file which you must first save to your disk and then double-click to extract the eight files.|
|The sample dataset was kindly provided by the Korean Institute of Tuberculosis (Dr. Sang Jae Kim).
Neuenschwander BE. Bayesian mixture analysis for tuberculin induration data. International Union Against Tuberculosis and Lung Disease, 2007, at http://www.tbrieder.org.
Link to the R Web site to download the R software
Link to the R Web site: http://lib.stat.cmu.edu/R/CRAN
Detailed descriptions on how to download R for various platforms are given. First go to R binaries. If you use Windows, go to Windows, then to base, and download R-2.4.1-win32.exe
Important Note for users: Jose Becerra from CDC kindly pointed out an incompatibility problem with newer versions of R: the script NONBCG.R (others not yet tested) results in an error message when used with the newest Version of R.
For installation of R, start the R-2.4.1-win32.exe.
The R icon should then appear on your desktop: double-click on it to start R. If everything works properly you should now be in R:. The R prompt is represented by the symbol >.
An example: type rnorm(5,10,2) at the R prompt. This should generate 5 normal random numbers with mean 10 and standard deviation 2.
Type y <- 2 and enter. “<-“ is used to assign values to a variable. Type y*y and enter.
Last update: October 1, 2010