Artificial Intelligence Technique in Privacy and Security of Patient Records for Healthcare Application

Abstract

Healthcare data analysis is increasingly critical today. Patient Healthcare Records (PHR) are essential for analyzing every detail pertaining to individual patients; therefore, the meticulous maintenance of all health information for patients is of paramount importance. A multitude of researchers are engaged in healthcare data analysis, ensuring the security and privacy of each Personal Health Record (PHR) analysis. Current methodologies predominantly utilize machine learning and artificial intelligence approaches; however, they fail to deliver precise accuracy results. Consequently, our paper introduces lightweight deep learning techniques grounded in cryptographic methods. Typically, image processing approaches include four fundamental analytical functions: pre-processing, segmentation, feature extraction, and classification methods. A Gaussian filter is employed to eradicate speckle noise seen in MRI or ultrasonic imaging of the brain. This technique involves substituting the noisy pixel, which follows a Gaussian distribution, with the average value of the adjacent or adjoining pixels. Preprocessing techniques facilitate the reduction of noise in all datasets. Our paper employs a Gaussian filter for pre-processing, utilizes principal component analysis for feature extraction, and applies enhanced homomorphic encryption through deep learning techniques to manage the datasets. Ultimately, our article evaluates its accuracy in relation to existing approaches.