Protecting Patient Data in AI/ML Models with Homomorphic Encryption in Hybrid Cloud Environments: Enabling Privacy-Preserving Analytics Without Decryption

Authors: Charan Shankar Kummarapurugu

DOI: https://doi.org/10.5281/zenodo.14059434

Short DOI: https://doi.org/g8qjqz

Country: USA

Full-text Research PDF File:   View   |   Download

Abstract: Healthcare applications increasingly rely on AI/ML models to analyze sensitive patient data, but the privacy of such data in cloud environments is a significant concern. This paper proposes a novel approach utilizing homomorphic encryp- tion to protect patient data while enabling privacy-preserving AI/ML analysis in hybrid cloud environments. By using this encryption method, we can process encrypted data without the need for decryption, thus ensuring data privacy and compliance with regulations such as HIPAA and GDPR. We explore the implementation challenges, performance trade-offs, and present an architecture to integrate homomorphic encryption in AI/ML workflows across hybrid clouds. The results show that this approach can effectively secure patient data while maintaining the accuracy and efficiency of AI models.

Keywords: Homomorphic Encryption, AI/ML Models, Hy- brid Cloud, Privacy-Preserving Analytics, Healthcare, Patient Data Security

Paper Id: 231472

Published On: 2021-06-03

Published In: Volume 9, Issue 3, May-June 2021

Share this