A Comprehensive Review of Convolutional Autoencoder with Dual-Key Transformer Network Based Smart E-Health Application for the Prediction of Tuberculosis Using Serverless Cloud Computing

Abstract

Tuberculosis (TB) remains one of the leading infectious diseases worldwide, requiring early and accurate diagnosis to reduce mortality and transmission. With the advancement of artificial intelligence and cloud computing, smart e-health systems have emerged as a promising solution for automated disease prediction. This paper presents a comprehensive review of a novel framework integrating Convolutional Autoencoders (CAE) with a Dual-Key Transformer Network deployed on a serverless cloud computing architecture for tuberculosis prediction. The CAE model effectively extracts latent features from medical imaging data such as chest X-rays, while the dual-key transformer enhances feature representation through attention mechanisms and secure key-based transformations. The integration of serverless computing platforms ensures scalability, cost efficiency, and real-time processing without infrastructure management overhead. The proposed approach demonstrates improved prediction accuracy, reduced latency, and enhanced data security compared to traditional machine learning and deep learning methods. Furthermore, the review highlights recent advancements in deep learning architectures, transformer-based models, and cloud-based healthcare systems. The study also identifies key challenges such as data privacy, model generalization, and computational constraints. Overall, this work provides insights into the potential of hybrid AI models combined with serverless cloud environments for next-generation intelligent healthcare applications.