A Novel Hybrid Model for Skin Disease Classification: Optimizing Feature Extraction and Sequential Learning for Accurate Dermatological Diagnosis

Skin diseases are among the most widespread health concerns globally, often leading to significant physical and psychological impacts.The visual similarity among different types of skin conditions poses a major challenge in early diagnosis and treatment.Additionally, manual diagnosis is subjective, time-consuming, and requires expert dermatological knowledge.To address these challenges, this study proposes a novel hybrid deep learning framework for automated and highly accurate skin disease classification using clinical images.The core innovation of the proposed method lies in the synergistic integration of multiple pre-trained Convolutional Neural Networks (CNNs)-ResNet50, DenseNet121, and MobileNetV2-for complementary and robust feature extraction.These features are concatenated into a unified feature vector, which is then fed into a Gated Recurrent Unit (GRU) network.This architecture effectively captures spatial and sequential dependencies in the image features, improving classification accuracy.This approach uniquely combines the spatial strengths of CNNs with the contextual learning capabilities of GRUs, a technique that remains underexplored in the field of dermatological image analysis.The model is trained and evaluated using a publicly available dataset titled "Skin Disease Dataset" from Kaggle, which includes high-resolution images across multiple skin disease categories.Prior to training, the dataset is subjected to rigorous preprocessing and augmentation techniques to enhance model generalization.The performance of the proposed hybrid model is compared against several state-of-the-art deep learning models using standard evaluation metrics such as accuracy, precision, recall, and F1-score.Experimental results show that the proposed method outperforms existing approaches, achieving a classification accuracy of 99.03%, demonstrating its effectiveness and robustness.This study contributes a significant advancement toward intelligent, computer-aided dermatological diagnosis and holds strong potential for deployment in real-time clinical settings.