A Privacy-Preserving Blockchain Framework for Secure Collaboration in Cloud-Based Applications

Abstract

The speed with which cloud computing has been used in cooperative healthcare applications has brought forth numerous concerns as far as data privacy, decentralized trust, and secure access control are concerned. The past systems, which were centralized, are highly susceptible to single points of failure and have a restricted ability to audit sensitive medical information. This paper presents a privacy-sensitive blockchain framework that enables cooperation among people in the early detection of heart disease to work efficiently. To guarantee the secrecy of sensitive identifiers of patients and the integrity of data, the framework employs enlightened cryptography techniques, including, but not limited to, AES-256-GCM encryption and salted SHA-256 hash. The main part of the system is the high-performance stacking ensemble machine learning model, which is composed of the following models: Random Forest, XG Boost, Light GBM, and Multi-Layer Perceptron (MLP). Possessing the ability to determine valid and suspicious access requests and effectively estimate the state of heart disease are its major functions. Smart contracts used to uphold an audit trail in an immutable fashion utilise a certified blockchain in which access decisions, audit metadata, and additional blockchain data are stored. To verify scalability and robustness, the framework is assessed using both synthetic and benchmark datasets (UCI Heart Disease). The performance of the ensemble model to improve the accuracy of an enhanced UCI Heart Disease dataset is demonstrated by experimental results showing 95% accuracy, 0.98 ROC-AUC, and better precision-recall than the performance of the individual classifiers. The results prove that the suggested framework provides a scalable, reliable, and regulation-conformant solution to medical collaboration in the cloud that will provide the best balance of top-level security and quality of clinical diagnosis.