Comparison of Three Anonymization Tools for a Health Fitness Study.

Area of Science:

• Data privacy
• Open science
• Health informatics

Background:

• Sharing personal data for research is crucial for open science.
• Anonymization is necessary to protect privacy but can reduce data utility.
• Evaluating anonymization tools requires assessing both privacy preservation and scientific validity.

Purpose of the Study:

• To evaluate the effectiveness of three data anonymization tools: ARX, SDV, and SynDiffix.
• To assess if anonymized datasets support the original scientific conclusions of a cardio-respiratory fitness study.
• To consider the practical burden on researchers using these anonymization tools.

Main Methods:

• A case study involving a dataset of Slovenian school children's fitness and commuting patterns (N=713).
• Application and evaluation of ARX, SDV, and SynDiffix anonymization tools.
• Comparison of scientific conclusions derived from original versus anonymized datasets.

Main Results:

• The study demonstrates that anonymization tools can successfully de-identify personal data while retaining scientific validity.
• Specific findings regarding the performance of ARX, SDV, and SynDiffix in preserving statistical significance and conclusions.
• Analysis of the usability and researcher burden associated with each tool.

Conclusions:

• Data anonymization is feasible for small, statistically sensitive datasets, enabling open science.
• The choice of anonymization tool impacts data utility and researcher workflow.
• This evaluation provides practical insights for researchers seeking to share sensitive health data.