Enhancing privacy and security in Federated learning protecting electronic health records data from adversarial attacks.
Saved in:
| Title: | Enhancing privacy and security in Federated learning protecting electronic health records data from adversarial attacks. |
|---|---|
| Authors: | Rajeswari, Burra Lalitha1 (AUTHOR) blalitharaji@gmail.com, Chakravarthy, A.S.N.1 (AUTHOR) |
| Source: | Informatics for Health & Social Care. 2026, Vol. 51 Issue 2, p151-168. 18p. |
| Subjects: | Federated learning, Electronic health records, Anonymity, Security management, Adversarial machine learning, Privacy |
| Abstract: | Federated Learning (FL) is a revolutionary approach to machine learning that allows multiple institutions, such as healthcare organizations, to collaboratively train models without sharing sensitive data directly. The problem involves protecting Electronic Health Records (EHRs) in FL from adversarial attacks while maintaining privacy. This requires ensuring secure data aggregation, anonymization, and robust model training across diverse healthcare institutions without compromising sensitive patient information. The objectives are to enhance privacy and security in FL for EHRs by implementing anonymization, adversarial pattern detection, context-aware aggregation, and robust model training. This ensures protection from adversarial attacks while maintaining compliance with privacy standards like HIPAA. Adaptive Weighted Median Filtering (AWMF) improves FL robustness by reducing noise, outliers, and adversarial attacks, ensuring accurate, privacy-preserving model training. The Cuckoo-based Deep Convolutional Long-Term Memory (CDC-LSTM) combines convolutional layers and memory networks to enhance FL, improving robustness against adversarial attacks while ensuring secure EHR data processing. Federated Privacy-Preserving Mesh Networks (F-PPMN) enhance FL by creating secure, decentralized communication channels, protecting EHRs from adversarial attacks and preserving privacy. Findings show that anonymized EHR attributes (90%-290%) slightly reduce data richness compared to original attributes (100%-300%) while maintaining privacy. This demonstrates a balance between data utility and confidentiality in FL and is implemented in Python Software. Future advancements include integrating federated encryption, multi-party computation, and improved adversarial attack detection, enhancing privacy and security in FL for EHRs while ensuring robust, accurate model performance across diverse healthcare environments. [ABSTRACT FROM AUTHOR] |
| Copyright of Informatics for Health & Social Care is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
Be the first to leave a comment!
