Analysis and Detection of Skin Diseases Using a Hybrid Deep Learning-Based Model for Dermatological Applications

Skin disease make up the most common medical conditions worldwide, and in order to avoid consequences, they must be diagnosed accurately and promptly. To identify and categorize ten different kinds of skin conditions, we provide a hybrid machine learning model in this study that combines CNN and LSTM networks. While the LSTM captures sequential dependencies and improves classification performance, the CNN component is excellent in extracting spatial characteristics from dermatological pictures. Using a robust dataset and optimized preprocessing techniques that includes augmentation of the original images, normalization of the images, transformation, and noise removal, the proposed hybrid model obtained a validation accuracy of 94.29%, and outperforms the existing model used for skin disease detection. This hybrid architecture demonstrates significant potential for aiding dermatological diagnostics by providing reliable and efficient automated solutions. Future work will explore integrating real-time applications and broader datasets to enhance the model’s generalizability.