<p>This paper presents a classification model for analyzing cervical cancer images, addressing one of the most prevalent cancers among women worldwide. Early detection is crucial for improving recovery rates and reducing mortality. The integration of artificial intelligence (AI) in medical diagnostics has shown promise in cervical cancer screening by enabling faster results, reducing dependency on specialists, and minimizing biases. While AI-based solutions exist, ongoing research aims to enhance accuracy and efficiency. Advancements in deep learning (DL) have facilitated the development of automated frameworks for medical image analysis, including cervical cancer detection. This research proposes a five-phase model: (1) pre-processing the dataset, (2) extracting features using pre-trained models, (3) optimizing feature selection with the Copula Entropy-based Golden Jackal Optimization (CE-based GJO) algorithm, (4) optimizing hyperparameters using the Sea Horse Optimizer (SHO), and (5) employing explainable AI (XAI) to identify key cytomorphological features in classification. The proposed model is trained on the SipakMed dataset, the largest publicly available cervical cancer dataset on Kaggle. Experimental results demonstrate the proposed model’s superior performance, achieving high precision (0.9985), specificity (0.9996), F-measure (0.9985), and accuracy (0.9985). It outperforms leading benchmark studies, highlighting its potential for precise cervical cancer diagnosis. Additionally, the model offers a secure, cost-effective, and efficient AI-driven solution for early detection and screening.</p>