Privacy-preserving data sharing via probabilistic modelling

TL;DR

This paper introduces a probabilistic modelling approach for generating privacy-preserving synthetic data, enabling accurate data sharing while maintaining privacy, demonstrated through epidemiological study results.

Contribution

It formulates private data release as a probabilistic modelling problem, incorporating prior knowledge to improve synthetic data quality.

Findings

Synthetic data reliably reproduces statistical discoveries

Method preserves privacy while maintaining data utility

Empirical validation in epidemiological research

Abstract

Differential privacy allows quantifying privacy loss resulting from accessing sensitive personal data. Repeated accesses to underlying data incur increasing loss. Releasing data as privacy-preserving synthetic data would avoid this limitation, but would leave open the problem of designing what kind of synthetic data. We propose formulating the problem of private data release through probabilistic modelling. This approach transforms the problem of designing the synthetic data into choosing a model for the data, allowing also including prior knowledge, which improves the quality of the synthetic data. We demonstrate empirically, in an epidemiological study, that statistical discoveries can be reliably reproduced from the synthetic data. We expect the method to have broad use in creating high-quality anonymized data twins of key data sets for research.