Scholarlink

In the era of digital transformation, privacy-preserving solutions are paramount for enabling secure and collaborative healthcare analytics. This paper presents a Federated Learning (FL) framework for fraud detection in healthcare, ensuring privacy preservation and compliance with regulatory standards. The proposed approach integrates differential privacy and secure aggregation to protect sensitive patient and provider data while enabling multi-stakeholder collaboration. Evaluation on a comprehensive healthcare dataset demonstrates the effectiveness of the framework, achieving a testing accuracy of 93.5%, a precision of 94.2%, and an F1-score of 93.1%. The model’s convergence is evident, with the global loss reducing from 1.345 in the initial round to 0.395 after 50 communication rounds. Furthermore, the impact of privacy budgets ( \(\epsilon \) ) on performance was analyzed, showing that even with a highly stringent budget ( \(\epsilon = 0.5\) ), the framework maintains an accuracy of 93.0% and an F1-score of 91.2%. Communication efficiency optimizations reduce data exchange by up to 90%, demonstrating scalability across distributed healthcare systems. These results highlight the potential of the FL framework to transform healthcare analytics, ensuring privacy, accuracy, and efficiency in a multi-stakeholder environment. This study sets a foundation for deploying FL in real-world healthcare settings, addressing critical challenges in data security and collaborative analytics.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Federated Learning for Privacy-Preserving Healthcare Analytics: A Novel Framework for Fraud Detection in Healthcare

  • 章节

  • 发表日期：

    2026年

查看全文
  • Md. Al Rafi,
  • Gourab Nicholas Rodrigues,
  • Md. Shahriar Mahmud Bhuiyan,
  • Md. Nazmul Hossain Mir,
  • Md. Nahid Hasan

摘要

In the era of digital transformation, privacy-preserving solutions are paramount for enabling secure and collaborative healthcare analytics. This paper presents a Federated Learning (FL) framework for fraud detection in healthcare, ensuring privacy preservation and compliance with regulatory standards. The proposed approach integrates differential privacy and secure aggregation to protect sensitive patient and provider data while enabling multi-stakeholder collaboration. Evaluation on a comprehensive healthcare dataset demonstrates the effectiveness of the framework, achieving a testing accuracy of 93.5%, a precision of 94.2%, and an F1-score of 93.1%. The model’s convergence is evident, with the global loss reducing from 1.345 in the initial round to 0.395 after 50 communication rounds. Furthermore, the impact of privacy budgets ( \(\epsilon \) ) on performance was analyzed, showing that even with a highly stringent budget ( \(\epsilon = 0.5\) ), the framework maintains an accuracy of 93.0% and an F1-score of 91.2%. Communication efficiency optimizations reduce data exchange by up to 90%, demonstrating scalability across distributed healthcare systems. These results highlight the potential of the FL framework to transform healthcare analytics, ensuring privacy, accuracy, and efficiency in a multi-stakeholder environment. This study sets a foundation for deploying FL in real-world healthcare settings, addressing critical challenges in data security and collaborative analytics.