Back

Data Anonymization for Open Science: A Case Study

Francis, P.; Jurak, G.; Leskosek, B.; Prasser, F.; Otte, K.

2024-12-16 health informatics
10.1101/2024.12.16.24319068 medRxiv
Show abstract

One of many challenges to open science is anonymization of personal data so that it may be shared. This paper presents a case study of the anonymization of a dataset containing cardio-respiratory fitness and commuting patterns for Slovenian school children. It evaluates three different anonymization tools, ARX, SDV, and SynDiffix. The fitness study was selected because its small size (N=713) and generally low statistical significance make it particularly challenging for data anonymization. Unlike most prior anonymization tool evaluations, this paper examines whether the scientific conclusions of the original study would have been supported by the anonymized datasets. It also considers the burden imposed on researchers using the tools both for data generation and data analysis.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.