Swin-PSO-SVM: A Novel Hybrid Model for Monkeypox Early Detection

A resurgent zoonotic disease called monkeypox presents serious public health issues, especially in locations with low resources where prompt and correct diagnosis is essential. Due to the health shortcomings in sensitivity, specificity, and adaptability to different clinical presentations, traditional diagnostic techniques frequently need to be revised. We introduce a novel hybrid Swin-PSO-SVM to improve the precision and dependability of monkeypox detection. The Swin-PSO-SVM model incorporates a Swin Transformer for complex feature extraction, Particle Swarm Optimization (PSO) for extracting the best features from complex feature extraction, and a Support Vector Machine (SVM) for accurate classification. The findings show how well the Swin-PSO-SVM model performs, attaining great diagnostic accuracy and resilience over various datasets and picture characteristics using two datasets: MSLD with two classes (monkeypox and others) and MSID with four classes(chickenpox, measles, monkeypox, and normal). Swin-PSO-SVM model has the highest accuracy of 95.556 and an F1 score of 95.569 on the MSLD dataset and a 96.429 accuracy and a 96.429 F1 score on the MSID dataset, outperforming several existing models. With an accuracy of 91.111 and an F1-score of 91.138, the experimental findings proved the model's exceptional performance and validated its dependability in real-world applications. The Swin-PSO-SVM provides a workable and understandable solution that can be easily applied in clinical settings and improves the generalizability of monkeypox detection. This paper contributes to worldwide efforts to limit monkeypox outbreaks through early and accurate identification.