Secure Stream Processing for Medical Data.

Area of Science:

• Computer Science
• Biomedical Engineering
• Cybersecurity

Background:

• Medical data ownership is increasingly complex, with sensitive information often processed by third-party cloud services without explicit consent.
• Growing concerns over data privacy and regulations like GDPR necessitate advanced privacy-preserving techniques for cloud-based data processing.
• Wearable devices generate continuous cardiac data, requiring secure and efficient cloud-based analysis.

Purpose of the Study:

• To present a proof-of-concept for a secure, streaming Internet of Things (IoT) architecture for cloud-based cardiac data processing.
• To demonstrate a method for enhancing data security without altering existing server-side application code.
• To evaluate the performance and privacy trade-offs of the proposed secure architecture.

Main Methods:

• Developed a streaming IoT architecture integrating trusted hardware with Apache Spark for secure cloud processing.
• Implemented a system capable of processing electrocardiogram (ECG) data from wearable sensors.
• Tested the architecture using a dataset of ECGs from healthy individuals undergoing physical activities (running, walking, biking).

Main Results:

• The proposed architecture successfully processed cardiac (ECG) data in the cloud with enhanced security guarantees.
• Security enhancements were achieved without requiring modifications to the server-side application code.
• Compared to standard Spark Streaming, the secure architecture resulted in a doubling of execution time.

Conclusions:

• The developed architecture offers a viable method for secure cloud processing of sensitive cardiac data from IoT devices.
• The integration of trusted hardware and Spark provides a robust framework for privacy-preserving data analysis.
• A trade-off exists between enhanced data privacy and computational performance, with execution time increasing by approximately 100%.