Abstract
A generic template for clinical trials simulations that are typically required by statisticians is developed. Realistic clinical trials data sets are created using a unifying model that allows general correlation structures for endpoint*timepoint data and nonnormal distributions (including time-to-event), and computationally efficient algorithms are presented. The model allows for patient dropout and noncompliance. A grid-enabled SAS-based system has been developed to implement this model; details are presented summarizing the system development. An example illustrating use of the system is given.
Original language | English |
---|---|
Pages (from-to) | 611-630 |
Number of pages | 20 |
Journal | Journal of Biopharmaceutical Statistics |
Volume | 18 |
Issue number | 4 |
DOIs | |
State | Published - Jul 2008 |
Keywords
- Binary data
- Correlation structure
- Grid computing
- Mixture data
- Non-compliance
- Non-normality
- Ordinal data
- Patient dropouts
- Survival data