tdm - A Tool for Therapeutic Drug Monitoring in R Using OpenBUGS

• Main Authors: Miao-Ting Chen, 陳妙婷
• Other Authors: Yung-jin Lee
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/87647640475863272094

碩士 === 高雄醫學大學 === 臨床藥學研究所碩士班 === 95 === In order to solve limitation of therapeutic drug monitoring (TDM), we used Bayesian MCMC integration algorithm, another Bayesian approach but not the same as the algorithm of minimizing an objective function, to predict individual PK/PD parameters with only limited clinical observation. Development tools applied in this study included OpenBUGS, Bayesian Using Gibbs Sampler (BUGS), as the core, R as the user interface and BRugs as connection between R and OpenBUGS. In this study, we not only built a tool but also explored some issues related to optimization of PK model settings. Effects of burn-in, update, initial value, the prior, Markov chain convergence and etc. were investigated. We named this package as tdm. Drug PK models we built in tdm were the first one applying to TDM using BUGS. After modifications, settings of burn-in, update, initial values and the prior were considered to be optimized. tdm runs on MS Windows OS. Seventeen drug PK models including one PK/PD model (warfarin) and sixteen PK models were built in tdm. It can be used to estimate individual PK/PD parameters with one or more observed values of one single subject, as well as multiple subjects at the same time. Furthermore, functions of dose adjustment were also provided. Finally, convergence diagnostic plots and summary information have been displayed. Also, we used simulated data to validate tdm. In conclusion, we have built tdm and applied Bayesian MCMC approach to prediction of individual PK parameters. And, it has been released at Nov. 2006 on the webs. More information is available from http://tdm.pkpd.org.tw/.