Cecilia Ritz Comparing prognostic markers for metastases in breast cancer using artificial neural networks Master Thesis Advisor: Patrik Edén Abstract: An important task of breast cancer treatment is to accurately diagnose the risk of metastasis. In this thesis artificial neural networks are used to compare the predictive power of conventional clinical observables and gene expression profiles of breast cancer tumours. Publicly available data from a previous study are used. In contrast to that study, we do not find that gene expression data outperforms conventional clinical observables. March 2003 LU TP 03-05