Abstract:
|
We review a set of parametric and non-parametric methods that are widely used in longitudinal data analysis, and illustrate their application with data from a published study of effect of alpha-difluoromethyl-ornithine (DFMO) on the growth of BT-20 human breast tumors in nude mice. Three parametric methods find significant difference between DFMO dose groups, while the non-parametric method shows that all groups follow a similar growth pattern, which provides reasonable support to fit a single mathematical model with different parameters. We find a correlation between model parameters and dosage, also another correlation between parameters and initial tumor volume is found in pool data. We conclude that by taking such two correlations into consideration, we can predict future tumor growth given initial tumor volume and dosage applied.
|