Performance Evaluation of two Prediction Models for Breast Cancer Metastasis Based on Data Mining Techniques, A Comparison Study

Abstract

Introduction: Defining the metastasis processes and what are the most effecting factors on improve the survival of patients and hopefully treating them. We aim to investigate and defining the factors predict breast cancer metastasis using data mining techniques. Data mining is the technique and tool of knowledge discovery from the big data. Nowadays data mining is spreading rapidly in several areas of research and business. In medicine, diagnosis of diseases is one of the fruitful and highly spreading filed of data mining.

Methods: There were 2025 usable records in ACECR Breast Disease Center’s data base after data preparation. We try to uncover the patterns that would help the prediction of metastasis factors using CHAID and Artificial Neural Network.

Results: After implementing mentioned algorithms, the tumor stage, surgery type and pathology results, the most important variables in metastasis prediction.

Conclusion: Comparing the algorithms execution revealed that, Artificial Neural Network, CHAID are convenient prediction models for breast cancer metastasis.