A diagnostic software tool for skin diseases with basic and weighted K-NN

Abstract

In the field of dermatology, to able to make differential diagnosis of erythemato-squamous diseases between each other accurately, is quite significant for the treatment of the disease. Especially the symptoms seen in the early stages of diseases in this group, may be very similar to each other. And this situation makes it difficult to determine accurate diagnosis for patients. Hence this study presents a data mining application for this problem in the medical field. Mentioned software tool in this study is trying to obtain correct diagnosis of erythemato-squamous diseases by using the basic and weighted K-NN algorithms on medical data. In this way, this paper presents a comparison between these two methods by evaluating and presenting the performances of them. Furthermore there is also a comparison between the Euclidean and Manhattan distance measures. © 2012 IEEE.