Secure Collaborative Platform for Health Care Research in an Open Environment: Perspective on Accountability in Access Control

Abstract

Background

With the recent use of IT in health care, a variety of eHealth data are increasingly being collected and stored by national health agencies. As these eHealth data can advance the modern health care system and make it smarter, many researchers want to use these data in their studies. However, using eHealth data brings about privacy and security concerns. The analytical environment that supports health care research must also consider many requirements. For these reasons, countries generally provide research platforms for health care, but some data providers (eg, patients) are still concerned about the security and privacy of their eHealth data. Thus, a more secure platform for health care research that guarantees the utility of eHealth data while focusing on its security and privacy is needed.

Objective

This study aims to implement a research platform for health care called the health care big data platform (HBDP), which is more secure than previous health care research platforms. The HBDP uses attribute-based encryption to achieve fine-grained access control and encryption of stored eHealth data in an open environment. Moreover, in the HBDP, platform administrators can perform the appropriate follow-up (eg, block illegal users) and monitoring through a private blockchain. In other words, the HBDP supports accountability in access control.

Methods

We first identified potential security threats in the health care domain. We then defined the security requirements to minimize the identified threats. In particular, the requirements were defined based on the security solutions used in existing health care research platforms. We then proposed the HBDP, which meets defined security requirements (ie, access control, encryption of stored eHealth data, and accountability). Finally, we implemented the HBDP to prove its feasibility.

Results

This study carried out case studies for illegal user detection via the implemented HBDP based on specific scenarios related to the threats. As a result, the platform detected illegal users appropriately via the security agent. Furthermore, in the empirical evaluation of massive data encryption (eg, 100,000 rows with 3 sensitive columns within 46 columns) for column-level encryption, full encryption after column-level encryption, and full decryption including column-level decryption, our approach achieved approximately 3 minutes, 1 minute, and 9 minutes, respectively. In the blockchain, average latencies and throughputs in 1Org with 2Peers reached approximately 18 seconds and 49 transactions per second (TPS) in read mode and approximately 4 seconds and 120 TPS in write mode in 300 TPS.

Conclusions

The HBDP enables fine-grained access control and secure storage of eHealth data via attribute-based encryption cryptography. It also provides nonrepudiation and accountability through the blockchain. Therefore, we consider that our proposal provides a sufficiently secure environment for the use of eHealth data in health care research.