Skip to Main Content
With the advancement of computer science; the database concept has been widely used in medical information system for processing large volumes of data. Symbolic and numeric data will define the need for new data analysis techniques and tools for knowledge discovery. Three popular algorithms for data mining which includes Bayesian Network (BN), C4.5 in Decision Tree (DT), and Back Propagation Neural Network (BPN) were evaluated in this paper. Two classes of dataset are used as testing data. The first dataset, Fine Needle Aspiration Cytology, is used to check whether the breast tumor is malignant. The second dataset, Tongue Diagnosis Image, is used to check whether upper GI is disorder. The result shows that BN had a good presentation in diagnosis ability. By using BN showed the accuracy about 94.6% in diagnosing breast tumor and 85.5% in upper GI disorder. The C4.5 learning algorithms in DT was able to explain diagnosis knowledge and rules. Its accuracy was 94.4% to diagnose a breast tumor, but it was just only 63.9% in upper GI disorder. The best performance among these three algorithms is BPN and its accuracy is 96.0% in diagnosing a breast tumor and 91.6% in diagnosing upper GI disorder, respectively.