The study compared different ways to predict when HIV-positive women might develop AIDS or die. Nonparametric models are more flexible but less efficient than parametric models, which make more assumptions. The researchers found that adding more data to the nonparametric model made it as efficient as the parametric models. The generalized gamma model was the best fit, while the exponential model had some errors. Adding more data can improve accuracy, but it may introduce bias if the model assumptions are wrong.