Author(s): Amirhossein Rahmani, Leila Farzaneh, Reza Khosravi and Niloofar Shadmehr

Abstract: The rapid digitization of healthcare information systems has introduced both opportunities for advanced analytics and challenges concerning the protection of sensitive patient data. In Iran, where the integration of electronic health records and clinical databases is expanding, concerns about cybersecurity, data breaches, and compliance with privacy regulations remain critical. This research investigates homomorphic encryption as a framework for secure data mining in Iranian healthcare systems, with a focus on its feasibility, efficiency, and accuracy in real-world applications. Using anonymized patient datasets from selected Iranian hospitals, predictive models such as logistic regression and decision trees were applied directly to encrypted data through the Brakerski-Gentry-Vaikuntanathan (BGV) scheme implemented in HElib. Comparative analysis with conventional AES-based workflows assessed encryption times, computational performance, and classification accuracy. The results revealed that homomorphic encryption preserved predictive accuracy with less than 2% loss compared to plaintext analysis, confirming that analytical validity is maintained while ensuring robust privacy. Although encryption and computation times were higher for homomorphic encryption, scalability testing demonstrated that the overhead became less significant with larger datasets, supporting its practicality for nationwide healthcare applications. The study concludes that homomorphic encryption can serve as a secure and efficient framework for privacy-preserving data mining in Iranian healthcare information systems. Practical recommendations include phased integration into hospital systems, investment in computing infrastructure, specialized training for healthcare IT personnel, and the formulation of national guidelines for privacy-preserving analytics. By adopting this approach, Iranian healthcare systems can enhance trust, regulatory compliance, and patient data security while enabling innovation in predictive analytics and decision support.

DOI: 10.33545/27076636.2025.v6.i2a.126

Pages: 162-166 | Views: 86 | Downloads: 39

Download Full Article: Click Here