Perturbed Gibbs Samplers for Generating Large-Scale Privacy-Safe Synthetic Health Data

By Yubin Park, Joydeep Ghosh, Mallikarjun Shankar. 2013

33 citations recorded by the source corpus.

This paper introduces a non-parametric data synthesizing algorithm to generate privacy-safe ``realistic but not real'' synthetic health data. Our goal is to provide a systematic mechanism that guarantees an adequate and controllable level of privacy while substantially improving on the utility of public use data, compared to current practices by CMS, OSHPD and other agencies. The proposed algorithm synthesizes artificial records while preserving the statistical characteristics of the original data to the extent possible. The risk from ``database linking attack'' is quantified by either an l-diversified or an ϵ-differentially perturbed data generation process. Moreover its algorithmic performance is optimized using Locality-Sensitive Hashing and parallel computation techniques to yield a linear-time algorithm that is suitable for Big Data Health applications. We synthesize a public Medicare claim dataset using the proposed algorithm, and demonstrate multiple data mining applications and statistical analyses using the data. The synthetic dataset delivers results that are substantially identical to those obtained from the original dataset, without revealing the actual records.

Subfields: Artificial Intelligence, Epidemiology, Management Science and Operations Research, Evaluation & Benchmarks, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jimeng Sun
2. Edward Choi
3. Walter F. Stewart
4. Brett K. Beaulieu-Jones
5. Casey S. Greene
6. Bill Howe
7. Sanjeev P. Bhavnani
8. Bernease Herman

In Epidemiology

1. Jimeng Sun
2. Jon D. Duke
3. Bradley A. Malin
4. Siddharth Biswal
5. Edward Choi
6. Walter F. Stewart

In Management Science and Operations Research

1. Brett K. Beaulieu-Jones
2. Jimeng Sun
3. Zhiwei Steven Wu
4. Casey S. Greene
5. Jon D. Duke
6. Bradley A. Malin
7. Siddharth Biswal
8. Edward Choi

In Evaluation & Benchmarks

1. Julia Stoyanovich
2. Jimeng Sun
3. Zhiwei Steven Wu
4. Jon D. Duke
5. Bradley A. Malin
6. Siddharth Biswal
7. Edward Choi
8. Walter F. Stewart

In Safety, Robustness & Interpretability

1. Jimeng Sun
2. Bernease Herman
3. Brett K. Beaulieu-Jones
4. Julia Stoyanovich
5. Bill Howe
6. Zhiwei Steven Wu
7. Jon D. Duke
8. Bradley A. Malin

Citations in the corpus

• Privacy-Preserving Generative Deep Neural Networks Support Clinical Data Sharing (2017, methodology): “Other methods aimed at performing this task generally fall into two groups: 1.) sampling methods with a quantifiable privacy risk [7], or 2.)”
• Generating Multi-label Discrete Patient Records using Generative Adversarial Networks (2017, methodology): “More recently, and most related to our work, a privacy-preserving patient data generator was proposed based on a perturbed Gibbs sampler (Park et al., 2013).”
• Synthetic Data for Social Good (2017, methodology): “Other recent approaches in privacy-preserving data generation include the work on plausible deniability in data synthesis [4], on perturbed Gibbs sampling for private data release [19, 18], and on sampling from differentially private copula functions [14].”
• Generating Multi-label Discrete Electronic Health Records using Generative Adversarial Networks (2017, methodology): “More recently, and most related to our work, a privacy-preserving patient data generator was proposed based on a perturbed Gibbs sampler (Park et al., 2013).”