A Correlation Blockchain Matrix Factorization to Enhance the Disease Prediction Accuracy and Security in IoT Medical Data

Abstract

An IoT software product’s reliability is the probability of the product working “correctly” under or over a given time. New opportunities are the result of expansion in the fast-paced Internet of Things (IoT) space. IoT technologies on the collected datasets improve disease progression technology, disease prediction, patient self-management and clinical intervention. To propose, the IoT with cipher block chaining in the traditional cryptographic operation mode will be used for cryptographic processing. Developing models for the supervised learning classification and security of imbalanced datasets is challenging, especially in the medical field. However, most real-time IoT datasets present most traditional machine learning algorithms challenging unbalanced datasets. Proposed a new framework for the Correlation Blockchain Matrix Factorization Classifier (CBMFC) related to comprehensive medical records. CBMFC uses a multiple class label machine learning that represents an independent population model based on disease meta functions such as profile age, group, or cognitive function keys. The Pairwise Coupling Multi-Class Classifier (PCMC) is used to prove the model’s correctness. This produces more comprehensive data in various machine learning environments, such as predictive classification, similar to real data performance. For the results of security analysis confirmation, the proposed IoT application model’s effectiveness can withstand various attacks, such as selected cryptographic attacks. In this proposed CBMFC system, classification accuracy, precision, recall, execution time and security matrix are used to evaluate performance.