Scholarlink

<p>The security and efficient management of healthcare data—especially in blood bank supply chains are of paramount importance due to the sensitive, diverse, and time-critical nature of the information involved. Existing approaches frequently fall short in balancing data protection, computational efficiency, and compliance with privacy regulations. This study introduces a robust, privacy-preserving framework that integrates AES-GCM encryption and hash-block storage to ensure secure cloud-based data handling. A novel component of this framework is the host-proof storage feature selector, which dynamically identifies sensitive features from healthcare datasets for secure cloud storage without compromising data usability. The framework employs the Banyan Tree Growth Optimization algorithm to fine-tune the hyperparameters of the XGBoost classifier, significantly enhancing prediction accuracy and minimizing processing time. To ensure trust and transparency in data retrieval, an Integrity Verification Block incorporating a Third-Party Auditor (TPA) is designed, using SVM-based feature matching to validate data authenticity within the cloud. Experimental evaluation on real-world healthcare datasets demonstrates the proposed system’s high effectiveness across multiple metrics. The BTGO-optimized XGBoost model achieved a classification accuracy of 99%, a reduction in error rate, and a 37% improvement in processing time compared to baseline models. Encryption latency averaged 0.23&#xa0;s, and integrity verification via TPA was completed within 0.12&#xa0;s. These results highlight the system’s ability to improve data reliability, security, and regulatory compliance while ensuring scalability and efficiency in cloud environments. Overall, the proposed model addresses critical limitations in existing solutions and offers a practical, secure, and high-performance approach to healthcare data management in real-world cloud-based applications. The framework achieved 99.72% classification accuracy, with AES-GCM encryption latency reduced to 0.23s and third-party verification performed in under 0.12s, validating the model’s real-time effectiveness in secure healthcare data management. The proposed model achieved a classification accuracy of 99.72%, outperforming baseline methods such as SVM and RF by over 6%. Compared to similar encryption-integrated models, our approach demonstrated 2–3× lower latency and real-time verifiability, confirming its suitability for cloud-based healthcare applications.</p>

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

Integrating machine learning and encryption for effective data management in blood bank supply chains

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月9日

查看全文
  • K. M. Kirupa Shankar,
  • V. Santhi

摘要

The security and efficient management of healthcare data—especially in blood bank supply chains are of paramount importance due to the sensitive, diverse, and time-critical nature of the information involved. Existing approaches frequently fall short in balancing data protection, computational efficiency, and compliance with privacy regulations. This study introduces a robust, privacy-preserving framework that integrates AES-GCM encryption and hash-block storage to ensure secure cloud-based data handling. A novel component of this framework is the host-proof storage feature selector, which dynamically identifies sensitive features from healthcare datasets for secure cloud storage without compromising data usability. The framework employs the Banyan Tree Growth Optimization algorithm to fine-tune the hyperparameters of the XGBoost classifier, significantly enhancing prediction accuracy and minimizing processing time. To ensure trust and transparency in data retrieval, an Integrity Verification Block incorporating a Third-Party Auditor (TPA) is designed, using SVM-based feature matching to validate data authenticity within the cloud. Experimental evaluation on real-world healthcare datasets demonstrates the proposed system’s high effectiveness across multiple metrics. The BTGO-optimized XGBoost model achieved a classification accuracy of 99%, a reduction in error rate, and a 37% improvement in processing time compared to baseline models. Encryption latency averaged 0.23 s, and integrity verification via TPA was completed within 0.12 s. These results highlight the system’s ability to improve data reliability, security, and regulatory compliance while ensuring scalability and efficiency in cloud environments. Overall, the proposed model addresses critical limitations in existing solutions and offers a practical, secure, and high-performance approach to healthcare data management in real-world cloud-based applications. The framework achieved 99.72% classification accuracy, with AES-GCM encryption latency reduced to 0.23s and third-party verification performed in under 0.12s, validating the model’s real-time effectiveness in secure healthcare data management. The proposed model achieved a classification accuracy of 99.72%, outperforming baseline methods such as SVM and RF by over 6%. Compared to similar encryption-integrated models, our approach demonstrated 2–3× lower latency and real-time verifiability, confirming its suitability for cloud-based healthcare applications.