On the Use of Estimated Tumor Marker Classifications in Tumor Diagnosis Prediction - A Case Study for Breast Cancer

In this paper we describe the use of tumor marker estimation models in the prediction of tumor diagnoses. In previous work we have identified classification models for tumor markers that can be used for estimating tumor marker values on the basis of standard blood parameters. These virtual tumor markers are now used in combination with standard blood parameters for learning classifiers that are used for predicting tumor diagnoses. Several data based modeling approaches implemented in HeuristicLab have been applied for identifying estimators for selected tumor markers and cancer diagnoses: Linear regression, k-nearest neighbor learning, artificial neural networks, and support vector machines (all optimized using evolutionary algorithms) as well as genetic programming. In the results section we summarize classification accuracies for breast cancer; we compare classification results achieved by models that use measured marker values as well as models that use virtual tumor markers.