How (not) to protect genomic data privacy in a distributed network: using trail re-identification to evaluate and design anonymity protection systems

By Bradley A. Malin, Latanya Sweeney. 2004

280 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Public Health, Environmental and Occupational Health, 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

In Public Health, Environmental and Occupational Health

In Safety, Robustness & Interpretability

Citations in the corpus

Privacy-Preserving Generative Deep Neural Networks Support Clinical Data Sharing (2017, background): “Numerous linkage and membership inference attacks on both biomedical datasets [6-13] and from machine learning models [14-16] have demonstrated the ability to re-identify participants or reveal participation in a study.”
Generating Multi-label Discrete Patient Records using Generative Adversarial Networks (2017, background): “…(Sweeney, 1997; El Emam et al., 2011a), diagnoses (Loukides et al., 2010), lab tests (Atreya et al., 2013), visits across healthcare providers (Malin and Sweeney, 2004), and genomic variants (Erlich and Narayanan, 2014)) To mitigate re-identiﬁcation vulnerabilities, researchers in the…”
A practical approach to achieve private medical record linkage in light of public resources (2013, background): “5 6 These efforts are often hindered by the fact that patients ’ records are scattered across , 7 so that, as programs expand in scale and variety, it is crucial to integrate disparate resources.”
How To Break Anonymity of the Netflix Prize Dataset (2006, background): “Famous re-identification attacks include de-anonymization of a Massachusetts hospital discharge database by joining it with with a public voter database [22], de-anonymization of individual DNA sequences [19], and privacy breaches caused by (ostensibly anonymized) AOL search data [12].”
The disclosure of diagnosis codes can breach research participants' privacy (2010, background): “17 It was recently demonstrated that parts of trails can be suppressed to thwart this attack.”
Link prediction by de-anonymization: How We Won the Kaggle Social Network Challenge (2011, background): “This is a daunting task: a long line of research has demonstrated the feasibility of de-anonymization on various datasets [27], [21], [13], [17], [25], [26], [14], [1] and theoretical evidence exists for the impossibility of de-anonymization of high-dimensional data [3].”
PATE-GAN: Generating Synthetic Data with Differential Privacy Guarantees (2018, background): “The most common way to mitigate the privacy risk of sharing sensitive records is to de-identify the records - but it is by now well-known that records that have been de-identified can be easily re-identified by linking them to other identifiable datasets [30; 13; 24; 22; 14].”
Generating Multi-label Discrete Electronic Health Records using Generative Adversarial Networks (2017, background): “The review process by legal departments and institutional review boards can take months, with no guarantee of access (Hodge Jr et al., 1999).”
Robust De-anonymization of Large Sparse Datasets (2008, unknown): “Unlike [25, 22, 14], our algorithm is robust.”