Mining Knowledge of the Patient Record: The Bayesian Classification to Predict and Detect Anomalies in Breast CancerŽ

Abstract

Abstract: knowledge management, data mining, and text mining techniques have been adopted in various successful biomedical applications in recent years. Data Mining (DM) is the most important subfields in knowledge management (KM). It has been proven that data mining can enhance the KM process with better knowledge. In this paper, we investigate the application of DM techniques for mining knowledge of the patient record. The patient record represents documents of the patients examinations and treatments. Data Mining is the process of mining or extracting information from a data set and transform it into an understandable structure for further use. We propose a methodology for mining medical knowledge based on the Bayesian Classification to predict and detect anomalies in breast cancer. We use the Naïve Bayes Algorithm to develop this methodology. We illustrate the knowledge mining process by real examples of medical field. We investigate through these illustrations how knowledge is better mined and thus, reused when applying concepts and techniques of Data Mining. On the other hand, we investigate the potential contribution of the Naive Bayesian Classification methodology as a reliable support in computer‑aided diagnosis of such events, using the well‑known Wisconsin Prognostic Breast Cancer dataset. Finally, we will demonstrate the suitability and ability of the Naive Bayes methodology in Classification/Prediction problems in breast cancer.