Privacy–Preserving Data Discovery in Distributed Health Networks with Dynamic Consent

A distributed health-data network that answers queries by aggregating results instead of data, so records never leave the institution. Scales to hundreds of sites and ten million records in thirty seconds.

Privacy–Preserving Data Discovery in Distributed Health Networks with Dynamic Consent
Years2023VenueRWTH Aachen — COMSYSKindM.Sc. thesisGrade1.0 (German scale, 1.0 best)Projectgitlab.com →PaperDownload PDF →

Abstract

Electronic health data is becoming increasingly important for clinical research and medical treatment, with more use cases emerging to improve public health. The quantity of data significantly impacts its usefulness in these applications, resulting in the necessity of sharing health data across facilities and international boundaries. However, sharing sensitive health information risks patients’ privacy due to increasing hacking incidents and data leakages. In addition, governmental regulations impede the process of organizational cooperation to build larger datasets and prohibit the reuse of existing data for evolving use cases. Recent research on dynamic consent has shown that the active involvement of patients fulfills the requirement of informed consent and positively impacts patients’ willingness to contribute their data. In this thesis, we adopt state-of-the-art techniques to develop a new distributed health data network for data discovery with dynamic consent that builds upon result aggregation rather than data aggregation to eliminate the need to share patients’ health data. The evaluation of our test system shows that we can scale the network to hundreds of institutions and process datasets with more than ten million records within thirty seconds.