Poisson process and clinical trial recruitment in CCTRN study

Nainan Hei, The University of Texas School of Public Health


Recruitment is one of most basic and critical steps in conducting clinical trials. The results from my research will help the hospital or clinical trial centers to model the recruiting process and thus recruit enough and appropriate patients in an efficient way. In this research, we have three studies: TIME (Timing in Myocardial infarction Evaluation), LateTIME (Later Timing in Myocardial infarction Evaluation) and FOCUS (First Mononuclear Cells injected in the US). In each study, I separated the data into training set (75%) and validation set (25%). In the test set, I built three models: Poisson, Gamma and Poisson-Gamma conjugate model. After building the model, I used the training set to estimate the model parameters (λ for Poisson model; &agr; and β for the later two models), and then estimated the recruitment rate for each week. I used the validation part to derive the probability of recruiting enough patients in the last ¼ period, and select the best model for my studies. The results showed that Poisson-Gamma Bayesian model is much better than the simple Poisson and Gamma model.

Subject Area


Recommended Citation

Hei, Nainan, "Poisson process and clinical trial recruitment in CCTRN study" (2014). Texas Medical Center Dissertations (via ProQuest). AAI1566325.