An Analysis of the Survivability in SEER Breast Cancer Data Using Association Rule Mining

摘要

Medical professionals need a reliable methodology to predict the survivability of patients with breast cancer. In this work, a classical association rule mining algorithm-Apriori was adopted for analyzing the related association relationship between medical attributes of records and the survivability of patients. The SEER Dataset was used in this research. After the dataset was preprocessed, 29606 records was obtained. Each record contains 17 breast cancer related attributes. Then apriori algorithm was applied in these preprocessed records, 326 association rules about 'survived' and 22 association rules about 'not survived' were obtained finally. These discovered association rules indicate that the attributes of EOD-Lymph Node Involv and SEER historic stage A play important roles in the survivability of patients after analyzed and compared.