KI-basierte Analyse von Krebsregisterdaten

Cancer registries systematically collect data on the occurrence, treatment, and progression of cancer in order to further improve the care of cancer patients. In the AI-CARE joint project funded by the German Federal Ministry of Health, in which cancer registries from eight federal states are involved, artificial intelligence (AI) methods for analyzing cancer registry data are being developed and compared with statistical methods. Particular attention at the German Research Center for Artificial Intelligence (DFKI) in Lübeck is paid to survival time analysis after cancer diagnosis. In a first experiment, a data set from the Schleswig-Holstein cancer registry was used to compare different models for survival time analysis of lung cancer patients and to determine the most important survival factors using AI explainability methods. The AI models and the statistical methods used provide interesting results for the risk assessment of patients diagnosed with lung cancer. The explainability methods used in the AI models confirmed the results of clinical studies with regard to survival factors.