OPTIMAL CONTROL POINT-BASED DEEP KRONECKER MOBILE NETWORK FOR PRIVACY-PROTECTED DATA CLASSIFICATION IN HEALTHCARE

The disease prediction system plays a vital role in the healthcare system. In order to lead a healthy life, disease prediction systems are essential as they help in the early prediction of risks in disease. Due to the enormous amount of patient data, the sharing of data becomes complex. Also, the sensitive nature of this information provides the way for security and privacy issues. Thus, protecting sensitive information in the healthcare system remains a major challenge. From this perspective, suitable security and protection measures are necessary for privacy-protected data classification. In this study, the Deep Kronecker Mobile Network (DKMN) for privacy-protected data classification is proposed, which combines the Deep Kronecker Network (DKN) with MobileNet. The DKMN is designed by merging the DKN with MobileNet. Firstly, the input healthcare data are considered and optimal control points are generated by the Chronological Drawer Algorithm (CDA). The CDA integrates chronological concepts with the DA. Following this, noise-Free Homomorphic Encryption (FHE) is applied using the Matrix Operation for Randomization and Encryption (MORE) algorithm. Subsequently, DKMN is employed for privacy-preserving data classification. Lastly, optimization of control points for effective privacy protection is conducted employing CDA. Furthermore, the DKMN obtained a higher accuracy of 91.7%, TNR of 92.9% and TPR of 90.8%.