Medical diagnosis support and accuracy improvement by application of total scoring from feature selection approach

Melanoma is the most deadly form of skin cancer. Early detection and successful treatment of this disease often is possible. The main goal of this paper is to present results of application of feature selection method to find the most important or all important features that characterize melanocytic spots on the skin and in this way defining of a new Total Dermatoscopy Score formula. Thus, it is possible to decrease dimensionality of that problem. Results gathered during research focus on about six from thirteen descriptive attributes which are the most relevant and are stated as core attributes. Based on these attributes a simple total scoring method could be applied to improve prediction (diagnosis) results, additionally also reducing complexity of problem. Results were acquired by application of six different machine learning algorithms and estimated using several evaluation measures.