Dermatological treatment and diagnosis depend heavily on the categorization of skin diseases. However, problems like the selection of features and streamlining frequently need to be solved to achieve excellent accuracy in classification. In this paper, we integrate Extensive Learning Machines (ELMs) with feature weighting optimization to present a unique strategy for effective skin disease categorization. Our approach attempts to improve the effectiveness of classification by giving characteristics the proper weights according to how important they are in differentiating among various skin conditions. The excellent generalization capabilities and quick training time of ELMs lead us to use them as the classification model. A customized optimization technique is used to optimize the weights of the characteristics extracted from skin disease photos in the suggested method. Test outcomes on a reference dataset show that our approach performs better than conventional categorization strategies. Our approach has been shown through experimental findings on a benchmark dataset to achieve greater accuracy in skin disease diagnosis and to perform better than conventional classification algorithms. Dermatologists may find that the suggested framework helps them diagnose skin diseases more quickly and accurately by increasing the efficacy and efficiency of the categorization process.