Cancer is the second most cause of death in Austria and around 38000 people are diagnosed with cancer each year . The goal of this paper is to analyze methods for evaluation of risk factors in order to parametrize a micro simulation model for cancer prevalence. The focus of this paper is on modeling the survival time. This is done by the methods of survival analysis and model selection. Firstly, the survival function is estimated by the Kaplan-Meier estimate. Afterwards, a Cox proportional hazards regression is performed with all possible sets of parameters. These models are tested by twos with the likelihood ratio test in order to compare them.
Another approach is the so-called Lasso method. This method puts a constraint on the sum of the absolute values of the regression coefficients and in most cases forces some of the coefficients to go to zero. The Akaike Information Criterion is also applied. All three methods are compared and the parameters which are supported, at least to a certain extent, by all of them are included in the estimation of the survival time of the prevalence model.