A Comprehensive Review of Blockchain-Based Hybrid Contextual-ATNet Approach for Daily Diabetes Management: Predicting Insulin Dosage for Improved Control

Abstract

Diabetes mellitus is a widespread chronic disease requiring continuous monitoring and precise insulin dosage management to prevent severe complications. Traditional treatment approaches often fail to provide personalized control due to the complex and dynamic nature of glucose-insulin interactions. The growing availability of wearable devices and health data has created opportunities for intelligent systems that can improve diabetes management. This paper presents a comprehensive review of a blockchain-integrated hybrid deep learning framework, focusing on the Contextual-ATNet architecture for insulin dosage prediction. The model combines contextual embeddings with attention-based temporal networks to capture complex physiological patterns from continuous glucose monitoring, dietary intake, and activity data. Blockchain technology enhances the system by ensuring secure, decentralized data management, enabling privacy-preserving data sharing and tamper-proof record keeping. Applications include real-time glucose prediction, personalized insulin recommendation, and remote patient monitoring. The integration of federated learning further improves model generalization without compromising data privacy. Empirical studies demonstrate improved prediction accuracy and extended forecasting horizons compared to traditional methods. However, challenges such as system integration, scalability, and real-world deployment remain. This review highlights the potential of combining deep learning and blockchain for developing secure, intelligent, and personalized diabetes management systems.