Early Detection of Skin Cancer Using Machine Learning Approach

Abstract

Over the past decade, a skin cancer detection has attracted a lot of research attention in the both medical and soft computing domains because of the increasing number of applications for early detection of skin diseases.  Strong evidence suggests that skin cancer is one of the top three deadly cancers brought on by Deoxyribo Nucleic Acid (DNA) damage is skin, which can be fatal. Cells begin to grow uncontrollably because of this damaged DNA, and they are currently expanding quickly. There have been several studies done on the automated detection of cancer in photographs of skin lesions. The existing studies failed to properly address the problem of cancer detection accuracy. Therefore, this study aimed to propose a framework that combined data from various inputs to improve the accuracy of skin cancer detection. The proposed approach uses feature importance mapping and mining to convert a multidata input into a single output classification model, which has advantages over statistical approaches. Five distinct machine learning algorithms and model prediction results were used to evaluate the recommended solution. The results showed that the suggested method can collect data, analyze it, and make predictions with high accuracy, ranging from 85 to 99.6%.