Automatic de-identification of textual documents in the electronic health record: a review of recent research

Abstract

BackgroundIn the United States, the Health Insurance Portability and Accountability Act (HIPAA) protects the confidentiality of patient data and requires the informed consent of the patient and approval of the Internal Review Board to use data for research purposes, but these requirements can be waived if data is de-identified. For clinical data to be considered de-identified, the HIPAA "Safe Harbor" technique requires 18 data elements (called PHI: Protected Health Information) to be removed. The de-identification of narrative text documents is often realized manually, and requires significant resources. Well aware of these issues, several authors have investigated automated de-identification of narrative text documents from the electronic health record, and a review of recent research in this domain is presented here.MethodsThis review focuses on recently published research (after 1995), and includes relevant publications from bibliographic queries in PubMed, conference proceedings, the ACM Digital Library, and interesting publications referenced in already included papers.ResultsThe literature search returned more than 200 publications. The majority focused only on structured data de-identification instead of narrative text, on image de-identification, or described manual de-identification, and were therefore excluded. Finally, 18 publications describing automated text de-identification were selected for detailed analysis of the architecture and methods used, the types of PHI detected and removed, the external resources used, and the types of clinical documents targeted. All text de-identification systems aimed to identify and remove person names, and many included other types of PHI. Most systems used only one or two specific clinical document types, and were mostly based on two different groups of methodologies: pattern matching and machine learning. Many systems combined both approaches for different types of PHI, but the majority relied only on pattern matching, rules, and dictionaries.ConclusionsIn general, methods based on dictionaries performed better with PHI that is rarely mentioned in clinical text, but are more difficult to generalize. Methods based on machine learning tend to perform better, especially with PHI that is not mentioned in the dictionaries used. Finally, the issues of anonymization, sufficient performance, and "over-scrubbing" are discussed in this publication.

When Patient Engagement and Research Ethics Collide: Lessons from a Dementia Forum

“…The majority of patients were in favor of data sharing among researchers, especially within the field of AD research, although support for the sharing of personal information decreased with a higher risk of re-identifying the data [47][48][49][50][51][52][53][54]. Trinidad and colleagues [55] also found that both current and prospective research participants are generally supportive of the sharing of non-identifiable data among researchers.…”

De-identification of health records using Anonym: Effectiveness and robustness across datasets

“…An overview of approaches to PHI de-identification is provided by Meystre et al [8]. From their analysis, they concluded that methods based on linguistic resources, such as dictionaries, tend to perform better with rarely mentioned PHIs.…”

Automatic Declassification of Textual Documents by Generalizing Sensitive Terms

“…Later the need to sanitize the unstructured documents had come into notice. This need is revealed in initiatives from DARPA [11] or the Consortium for Healthcare Informatics Research (CHIR) [12] which aim at building new methods and tools for declassification of confidential documents. In the structured documents the structure itself provides the key to identify sensitive terms.…”