Securing cloud-based medical data: an optimal dual kernal support vector approach for enhanced EHR management

Abstract

Cloud computing is one of the advanced technologies to process rapidly growing data. At the same instant, the necessity of storage space for the voluminous digital medical data has been amplified thanks to the mounting electronic health records. It influences the employment of cloud outsourcing methodology. Data outsourced to the cloud space must be highly secured. For this, the paper presents a DKS-CWH algorithm that is based on a dual kernal support vector (DKS) and crossover-based wild horse optimization algorithm. In this paper, the input grayscale images are gathered from the medical MINST dataset which includes 58,954 images comprising six classes of CXR (chest X-ray), breast MRI, abdomen CT, chest CT, hand (hand X-ray), and head CT. The classification and feature extraction processes are performed at the cloud layer using the DKS-CWH algorithm. The hyperparameters of the DKS approach are optimized with the crossover-based WHO algorithm. The performance evaluation involves analyzing its effectiveness according to prominent metrics such as precision, accuracy, recall, and F1-score and comparing the outputs with the other competent methods. The results showed the DKS-CWH model offered robust performance with 97% accuracy.