De-identification of electronic health record using neural network

Abstract

According to a recent study, around 99% of hospitals across the US now use electronic health record systems (EHRs). One of the most common types of EHR is the unstructured textual data, and unlocking hidden details from this data is critical for improving current medical practices and research endeavors. However, these textual data contain sensitive information, which could compromise our privacy. Therefore, medical textual data cannot be released publicly without undergoing any privacy-protective measures. De-identification is a process of detecting and removing all sensitive information present in EHRs, and it is a necessary step towards privacy-preserving EHR data sharing. Over the last decade, there have been several proposals to de-identify textual data using manual, rule-based, and machine learning methods. In this article, we propose new methods to de-identify textual data based on the self-attention mechanism and stacked Recurrent Neural Network. To the best of our knowledge, we are the first to employ these techniques. Experimental results on three different datasets show that our model performs better than all state-of-the-art mechanism irrespective of the dataset. Additionally, our proposed method is significantly faster than the existing techniques. Finally, we introduced three utility metrics to judge the quality of the de-identified data.

Clinical Notes De-Identification: Scoping Recent Benchmarks for n2c2 Datasets

“…However, because of the different evaluation methods, this improvement cannot be taken at face value. Studies used different evaluation methods, from token-based binary [1,6,7,8] classification to entity classification based on HIPAA PHI [5,9].…”

“…In terms of methods, all the studies were deep learning-based, and invariably used bidirectional long-short term memory (Bi-LSTM), Conditional Random Fields (CRF) and Gated recurrent units (GRU). With these base methods, studies developed multiple innovative ensemble methods through voting mechanisms [3], stacking [3,7] and novel attention mechanisms based on transformer models [7,8], and use of rule-based methods and dictionaries [1,2,9].…”

“…In addition, there are multiple combinations of hyper-parameters, and optimization processes were never reported. Advancement of scientific knowledge depends on full disclosure of such information, but only two studies provided a code repository [7,9] of their work.…”

The OpenDeID corpus for patient de-identification

“…Unstructured electronic health records (EHRs) such as discharge summaries, encounter notes, pathology reports and radiology reports are valuable sources of information for undertaking basic, clinical and translational research 1 – 3 . Often, researchers are required to share or access EHRs in a de-identified state to protect the privacy of patients.…”

“…Over the years, several studies have been conducted on de-identification of EHRs using manual, rule based and machine learning methods. Automated de-identification methods can be employed to replace manual processes 1 , 7 – 11 . These methods, especially deep learning-based methods, often require a large corpus for accurate de-identification.…”