A Comprehensive Framework for Machine Learning-Based Threat Intelligence in Health Information Systems

Abstract

This research work provides a comprehensive architecture of Machine Learning based threat intelligence particularly for Health Information System (HIS). The number of cybersecurity threats executed by healthcare companies is even higher since healthcare organizations continue to introduce digitized data into medical data. This work employs complex machine learning techniques from the MIMIC-III Critical Care Database to develop a practical threat identification and mitigation system. In this case, the strategy of analysis involves selection of data, data processing, modeling and real time dangers identification considering both supervised and unsupervised learning. The results reveal that the proposed framework covers high performance indicators such as: accuracy that equals 97.92%, and the level of precision and recall which also equal 90% ROC AUC has reached 0.94. These results demonstrate that the framework can identify and categorise cybersecurity risks in systems of health information on a regular basis. It not only increases threat perception but also makes the system internally valuable for healthcare IT professionals since it contains real-time monitoring and anomaly detection functionality. Therefore, this study stands in support of the ongoing efforts to enhance the security of the healthcare bodies on the use of policies on cybersecurity so as to ensure the protection of individual patient’s information against new forms of threats.