Privacy-Preserving Data Sharing for Health Data Mining

点击翻译按钮获取中文摘要

Health data mining is the process of extracting useful knowledge from a large volume of health data in an efficient and scalable manner. An effective data sharing system plays an important role in supporting health data mining. The emergence of cloud computing has significantly improved the potential of sharing health data due to its powerful computation capability, flexible service composability, high availability, and low maintenance. In particular, the new paradigm of Data-as-a-Service (DaaS) in cloud computing is to provide flexible, on-demand data service to data recipients regardless of their devices, platforms, times, or locations, while charging only for what they use. These desirable properties make cloud computing a natural choice for the next generation of healthcare information infrastructure. Yet, the major obstacle to adopting this technology in the healthcare sector is a lack of trust in sufficient privacy protection. Many privacy-enhancing technologies (PETs) have been developed in the last decade, but most have not been effectively utilized in real-life health information systems due to the following problems: First, most PETs do not consider the information needs of health data mining. Second, they are designed for general data-sharing scenarios and are not flexible for accommodating the dynamic data-sharing scenarios in the healthcare sector. Third, they do not follow the standard communication protocols and data formats required in the healthcare sector. To remove the aforementioned obstacles, the objective of the proposed research is to develop a privacy-preserving DaaS-based health data sharing system for supporting health data mining. The proposed platform will enable health information custodians (HICs), such as hospitals, clinics, and labs, to securely share and integrate their patient-specific data, while the health data miners (HDMs), such as health professionals in other health agencies and researchers in pharmaceutical companies, can still effectively retrieve their required information and perform their anticipated operations and research activities.