MASTERÂ Assignment
Sub-quadratic Privacy Preserving cohort selection
Type : Master M-CS
Period: Mar, 2023 - Aug, 2023
Student : Gansel, A.X.G. (Antoine, Student M-CS)
Date Final project: August 16, 2023
Supervisors:
Abstract:
In this work, we present a Privacy Preserving Cohort Selection (PPCS) protocol for vertically partitioned data running in sub-quadratic time. Cohort selection is used in case-control studies to match a control group in a distant database, given the knowl- edge of a test group. Such studies allow one to efficiently put in evidence the effect of a variable (e.g. a medicine) on a situation (e.g. a disease) and are thus of significant importance in the medical field. By providing a tool to easily and efficiently respect the privacy of test subjects in such studies, we aim at mitigating concerns that would naturally arise when processing the data. In this work, we aspire to bridge a gap in the literature that mainly focused on PPCS for horizontally partitioned data until now, as well as to improve on the previous result running in quadratic time. In the follow- ing, we formally prove the privacy of our solution and show it achieves a complexity of O(n log2 (n)2 ). We show that it results in a concrete improvement on the result of previous research considering cohort selections at a European level, and we elabo- rate on the impact of the bandwidth bottleneck on real case executions of our protocol.