This paper examines how artificial intelligence (AI) and deep learning (DL) affect medical diagnostics by automating disease prediction with image recognition. Image recognition tools, such as convolutional neural networks (CNNs), Vision Transformers (VITs), and Generative Adversarial Networks (GANs), find useful patterns in medical images like X-rays, MRIs, CT scans, and retinal fundus images. Recently, models like EfficientNet and ResNet have shown good accuracy in predicting pneumonia. Meanwhile, GhostNet and Bi-DenseNet have been more effective in predicting eye diseases. Additionally, explainable AI (XAI) techniques like Grad-CAM and Saliency Maps offer visual methods to understand how models make decisions. This improves transparency in these algorithms and builds trust in clinical decision-making. While past and current projects highlight successes, challenges such as data inconsistency, limited interpretation, and integration into clinical workflows still exist. Looking ahead, the focus will shift to developing explainable deep learning, multi-model, and federated learning frameworks. The aim is to create reliable systems that support scalability, analysis, and ethical use of future AI algorithms. Overall, AI and DL, through image recognition, mark an important advancement toward precision medicine and better healthcare systems.

This paper reviews cutting-edge technology on artificial intelligence (AI) and deep learning (DL) in medical disease prediction via image classification. Models have used deep learning models such as convolutional neural networks (CNN), generative adversarial networks (GAN), and vision transformers (Vit) on brain, lung, eye, and accuracy and explicability in transparency. Future work may necessitate the enhancement of privacy, scale, and opportunity. Multi-modal learning and federated training with a continued agenda skin diseases that have reported high accuracy in diagnosis [1]– [4]. Nonetheless, [5], [6] clinicians are challenged with the balance of datasets, low generalizability, or poor interpretability in clinical use.The study proposes a hybridized framework of CNN, transformer, and GAN-based augmentation tools in a suitable magnitude with explainable AI (XAI) visualisation to improve for ethical validation could improve clinical assurance of the implementation of medical AI and for AI demands.

[1]N. Shilpa, W. A. Banu, and P. B. Metre, “Revolutionizing Pneumonia Diagnosis: AI-Driven Deep Learning Framework for Automated Detection from Chest X-Rays,” IEEE Access, vol. 12, pp. 171601–171612, 2024. [2]D. Sharma, A. K. Gupta, and P. Jain, “Medical Image Processing, Disease Prediction and Report Summarization Using Generative Adversarial Networks and AIML,” IEEE Access, 2024. [3]M. Fu, C. Tantithamthavorn, and T. Le, “DAVit: A Domain-Adapted Vision Transformer for Automated Pneumonia Detection and Explanation Using Chest X- Ray Images,” IEEE Access, vol. 13, pp. 103033–103050, 2025. [4]P. Kaushik, A. Sharma, and R. Singh, “PneumoAI: Redefining Accuracy in Pneumonia Detection Using Advanced Machine Learning,” IEEE Access, 2024. [5]D. Jeya Mala, M. Chattopadhyay, P. Mukhopadhyay, and R. Sinha, “Visualizing UNet Decisions: An Explainable AI Perspective for Brain MRI Segmentation,” IEEE Access, vol. 13, pp. 1–10, 2024. [6]R. Chen, E. Song, and Z. Zhou, “A Lightweight Neural Network for Ocular Surface Disease Recognition,” in Proc. IEEE ICAICE, pp. 671–677, 2024. [7]S. Ahmad, N. Li, and F. Wang, “Deep Learning in Skin Disease Image Recognition: A Review,” IEEE Access, 2024. [8]K. Vayadande, M. Patel, and P. Thakur, “Ocular Disease Recognition Using Deep Learning,” IEEE Access, vol. 10, pp. 150220–150229, 2022. [9]S. Iqbal, N. Khan, and Z. Ahmed, “Fusion of Textural and Visual Information for Medical Image Modality Retrieval Using Deep Learning-Based Feature Engineering,” IEEE Access, vol. 11, pp. 45387–45398, 2023. H. Singh, R. Verma, and S. Jain, “Ocular Disease Detection via Deep Learning Models,” IEEE Access, vol. 13, pp. 23422–23435, 2025.