Application of a Unified Medical Data Miner (UMDM) for Prediction, Classification, Interpretation and Visualization on Medical Datasets: The Diabetes Dataset Case

Medical datasets hold huge number of records about the patients, the doctors and the diseases. The extraction of useful information which will provide knowledge in decision making process for the diagnosis and treatment of the diseases are becoming increasingly determinant. Knowledge Discovery and data mining make use of Artificial Intelligence (AI) algorithms which are applied to discover hidden patterns and relations in complex datasets using intelligent agents. The existing data mining algorithms and techniques are designed to solve the individual problems, such as classification or clustering. Up till now, no unifying theory is developed. Among the different algorithms in data mining for prediction, classification, interpretation and visualization, ‘k-means clustering’, ‘Decision Trees (C4.5)’, ‘Neural Network (NNs)’ and ‘Data Visualization (2D or 3D scattered graphs)’ algorithms are frequently utilized in data mining tools. The choice of the algorithm depends on the intended use of extracted knowledge. In this paper, the mentioned algorithms are unified into a tool, called Unified Medical Data Miner (UMDM) that will enable prediction, classification, interpretation and visualization on a diabetes dataset.