PDx-Pop 4.20 Update Now Available

PDx-Pop 4.20 Update December 21, 2010 William J. Bachman, Ph.D. Updating PDx-Pop 4.x to PDx-Pop 4.20: 1. Quit the PDx-Pop program if currently running. 2. Copy to a backup folder any of the files with identical names in your PDx-Pop installation directory (if you installed PDx-Pop to the default directory, the directory is c:\pdxpop4) as those found in the ftp folder (ftp://ftp.globomaxnm.com/Public/pdxpop/PDx-Pop4.2/). 3. Download all the files found at ftp://ftp.globomaxnm.com/Public/pdxpop/PDx-Pop4.2/ <ftp://ftp.globomaxnm.com/Public/pdxpop/PDx-Pop4.2/> to your PDx-Pop 4 installation directory. Move the files 501.csv and wamexample.ctl to the example1 directory under the installation directory. 4. Start the program and you are now ready to use PDx-Pop 4.20. Update 4.20 is cumulative with update 4.10 (it contains the updates found in 4.10). There is no need to precede update 4.20 with update 4.10 if you have not already updated to 4.10. Software Bugs fixed in PDx-Pop 4.20: 1. A bug was corrected that affected all runs that used multiple simultaneous threads including selecting multiple CPU's from the Model/Run Tab or Evaluation Methods that used multiple simultaneous threads by default including Multiple MCMC Runs and Initial Parameters Variation. If the control files contained multiple $TABLE records, the runs would fail and PDx-Pop would "hang" (the program would stop functioning and could only be closed from the Windows Task manager). The bug fix corrects the bug and also limits the number of multiple simultaneous threads to the total number of CPU's available on the system to prevent excess memory use which could lead to skipped runs. 2. Bugs were fixed that caused Bootstrap, Multiple MCMC chains, and Initial Parameter Variation evaluation methods to fail on Linux and Mac OS X. New features have been added to PDx-Pop 4 in the new updated version, PDx-Pop 4.20: Wald Approximation Method (WAM) Analysis Introduction to WAM Analysis The Wald Approximation Method (WAM) as implemented in PDx-Pop is based upon the publication by KG Kowalski and MM Hutmacher, "Efficient Screening of Covariates in Population Models Using Wald's Approximation to the Likelihood Ratio Test.", JPP 2001;28:253-275. The WAM is an efficient covariate search algorithm that exploits information contained in a full model fit (all covariates included simultaneously) to guide selection of competing reduced models for evaluation. The main goal of any covariate search algorithm is to find a parsimonious model for prediction. The WAM generally requires fewer model runs than stepwise procedures to identify a final reduced model and provides a set of competing parsimonious models. See the ReleaseNotes.pdf file for more information.