Secure Cloud-Based Diabetes Classification with Machine Learning

With the continuous advancement of medical devices, a substantial amount of data is generated within the healthcare environment daily. Consequently, cloud computing technology is being employed to securely store, process, and manage these vast volumes of data, safeguarding them from potential attacks. This paper focuses on diabetes classification using machine learning (ML) algorithms with a secured cloud computing environment by combining the Elliptic Curve Cryptography (ECC) and the Advanced Encryption Standard (AES) method to ensure authentication and data integrity. Additionally, we performed a comparative analysis of diabetes prediction classification using five diverse ML algorithms. The results show that the random forest (RF) classification method achieves the best results among K-Nearest Neighbors (KNN), Naive Bayes (NB), Diction tree (DT), and support vector machine (SVM) classifiers, with an accuracy of 90.4%, as well as the evaluation of security in the cloud involves analyzing the encryption and decryption times for records of different file sizes, ranging from 20 KB to 50 KB. The results obtained demonstrate that the proposed system surpasses other state-of-the-art systems in terms of performance. Therefore, the proposed model successfully establishes an efficient remote health monitoring system that relies on secure data storage and management on cloud-based platforms.