Machine Learning-Driven Assessment and Security Enhancement for Electronic Health Record Systems

Abstract

The digitalized patient-centric system, the Electronic Health Record (EHR), is a platform where comprehensive health information is stored, managed, and accessed electronically. The primary findings of this study aim to secure sensitive patient data and increase overall system resilience by demonstrating that machine learning can evaluate vulnerabilities and improve the security of Electronic Health Record (EHR) systems. This research examines the prospects of incorporating machine learning-driven assessment tools and safety improvements in EHRs to enhance data protection in the healthcare industry. The proposed method utilizes the implementation of machine learning classifiers, specifically the XGBoost and LightGBM models. These classifiers are employed to enhance various aspects of the system, such as data protection and security, within the framework of EHRs. The study emphasizes the efficiency of these machine learning classifiers in ensuring that EHR systems are secure enough to deal with any problem that may occur due to threats posed by external factors or hackers. The findings reveal that the XGBoost model always has outstanding performance, with a near-perfect Receiver Operating Characteristic Curve (ROC) having an AUC equal to 1.00, indicating close to perfect accuracy in distinguishing positive from negative cases. Similarly, LightGBM has a perfect ROC curve as well. Therefore, its performance would be considered flawless. Consequently, future developments could lead to sophisticated machine learning models besides those that have already been developed. Improving data storage through encryption and building safer communication protocols should also be considered to make these systems withstand new security problems. Thus, this study contributes to the existing literature on applying technology to safeguard vulnerable medical records while fostering a safe and efficient healthcare ecosystem.

Ähnliche Arbeiten