Research and Analysis of Healthcare Data Breach Risk Prediction in the US Based on Interpretable Machine Learning

¹ School of Automation and Intelligent Sensing, Shanghai Jiao Tong University, Shanghai, China
² Information School, University of Washington, Seattle, Washington, The United States
³ College of Arts and Science, Syracuse University, Syracuse, New York, The United States

^* Corresponding author’s email: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

This research is based on the Office for Civil Rights at the Department of Health and Human Services (OCR-HHS) breach reports (2019-2025) to build an interpretable machine learning model that forecasts incidents with a high impact ( ≥ 100,000 people) and likelihood of a ransomware. They include entity type, breach method and the location of compromised information. A comparative analysis was made between the logistic regression and random forest models to provide transparency and accuracy, calculate the calibration analysis, and feature importance analysis. The anticipated benefits are actionable tiered controls risk scores, improved incident preparedness, and governance decision support of healthcare cybersecurity. Research is based only on publicly available aggregated statistics, but not on patient-related data, which meets professional and regulatory ethics. Findings indicate that the two models are better than random baselines and they can provide noteworthy early-warning information; however, overall, the discriminative ability is low. Critical elements - attack vectors and information location - provide consumable results even to operational security planning. Generally, the findings have shown that interpretable predictions that are data driven can be feasible in reinforcing proactive cybersecurity governance within the healthcare industry.