Modern estimation methods
In this chapter we discuss methods to estimate biased regression coefficients, which lead to better predictions than those obtained with traditional methods. These modern estimation methods include uniform shrinkage methods (heuristic or bootstrap based) and penalized maximum likelihood methods (with various forms of penalty, including the “Lasso”). We illustrate the application of these methods with a data set of 785 patients from the GUSTO-I case study. It appears that rather advanced procedures can now readily be performed with modern software.