Medical Data Mining :An Expert Diagnostic For Dermatological Diseases

Data mining is becoming fore front in the healthcare in- dustry today, with the key role that it plays in the prediction of diseases based on collated data. Medi- cal diagnosis is an important but complicated task that should be performed with great accuracy and efficiency. Various studies prove that the diagnosis of a single patient can differ significantly if examined by dif- ferent physicians or by the same physician at different times. Today, automated medical analysis help doctors to diagnose and predict diseases, at a very fast pace. This study addresses dermatological diseases which are largely neglected, but may even prove fatal if left unattended.Medical dataset used for this work contain 230 instances with 22 attributes. Weka is built in software tool for data mining. Five classification algorithms used are J48 (Decision tree), Naive Bayes' (NB), Multilayer perception (Artificial Neural Network), ZeroR (Rule based) and Multiclass classifier (Support Vector Machine). Prediction of dermatological diseases is very difficult because of the large number of similar disease presentations. In this paper we have experimented on data gathered from the southern part of Kerala, India. The GUI, developed in Java, reveals the chances of different dermatological disease and also finds out the probabilities of occurrence of each disease.Research Scholar, Karpagam University, Coimbatore, India. Tamil Nadu, India. 'Dean, Sri Ramakrishna Institute of Technology, Coimbatore, Tamil Nadu, India. 2Asst Professor, Dept of Dermatology, Govt Medical College, Kottayam, Kerala, India.