An intelligent framework for securing patients’ electronic health records using anomaly based detection technique

Electronic health records (HERs) are highly priced in the dark web. The dark web is an illegal market for the sale of stolen health records and other health services, among others. This has led to the exploitation of EHRs by malicious application programs (malware), privileged health workers, and external aggressors (hackers). Therefore, patients’ .privacy and security attributes are threatened. To curb the menace of malware, hackers and privileged workers, this paper has developed a framework using machine learning (ML) tools. Random forest (RF) algorithm is used to train base classifiers such as Decision Tree (DT), K-Nearest Neighbor (KNN), Gaussian Naïve Bayes (GNB), and Support Vector Machine (SVM). These models learn normal patterns in NSL-KDD dataset, acquire experience, and use that.knowledge to classify malware from normal applications in the cloud. The models were evaluated for generality and performance and had accuracies of 98% (RF), 98% (DT) and 96% (KNN). The use of this framework in electronic health systems (EHS) will deter mischief makers on the web from successfully stealing electronic health records (EHRs) and trading with them in the dark web. The effort of this work will reduce privacy threats, increase patients’ trust, and instill confidence in data sharing among hospitals.