Protected Health Information filter (Philter): accurately and securely de-identifying free-text clinical notes

Abstract

There is a great and growing need to ascertain what exactly is the state of a patient, in terms of disease progression, actual care practices, pathology, adverse events, and much more, beyond the paucity of data available in structured medical record data. Ascertaining these harder-to-reach data elements is now critical for the accurate phenotyping of complex traits, detection of adverse outcomes, efficacy of off-label drug use, and longitudinal patient surveillance. Clinical notes often contain the most detailed and relevant digital information about individual patients, the nuances of their diseases, the treatment strategies selected by physicians, and the resulting outcomes. However, notes remain largely unused for research because they contain Protected Health Information (PHI), which is synonymous with individually identifying data. Previous clinical note de-identification approaches have been rigid and still too inaccurate to see any substantial real-world use, primarily because they have been trained with too small medical text corpora. To build a new de-identification tool, we created the largest manually annotated clinical note corpus for PHI and develop a customizable open-source de-identification software called Philter ("Protected Health Information filter"). Here we describe the design and evaluation of Philter, and show how it offers substantial real-world improvements over prior methods.

Building a Best-in-Class Automated De-identification Tool for Electronic Health Records Through Ensemble Learning

“…We compared the performance of our approach on the 2014 I2B2 test dataset with six other established de-identification tools: Scrubber [40], Physionet [41], Philter [22], MIST [42] and a model proposed by Dernoncourt et al [27] that blends CRFs and artificial neural networks (ANNs) .…”

“…We compared the performance of our approach on the 2014 I2B2 test set with six other established de-identification tools: the method proposed by Dernoncourt et al that blends conditional random fields (CRFs) and artificial neural networks (ANNs) 18 , Scrubber 19 , Physionet 8 , Philter 20 , MIST 21 and NeuroNER 22…”

“…Firstly, we cite the conditional random field and artificial neural network approach (CRF+ANN) 18 scores against the group A entities (HIPAA only) as reported in their paper. We also directly report the results for Scrubber, Physionet, and Philter from prior publications 20 without performing an empirical analysis because the dataset (2014 I2B2) and the set of PII entities is the same as that used in our investigation. We trained MIST using sentences from the I2B2 training corpus (see Supplementary Methods and Supplementary Table 3 ).…”

“…Rule-based systems 19,20,30–32 use pattern matching rules, regular expressions, dictionary and public database lookups to identify PII elements. These are simple to implement and usually deterministic; however, these systems have several drawbacks.…”

Natural Language Processing–Based Virtual Cofacilitator for Online Cancer Support Groups: Protocol for an Algorithm Development and Validation Study

“…The text data were deidentified to ensure confidentiality, using a risk-based approach in 2 steps. The first step used an open-source clinical text deidentification library (philter-ucsf [ 32 ]) to replace identifiers such as name, address, sex, age or year mention, and health care organization (hospital) names with asterisks, preserving word length. The second step involved human review and deidentification of any identifiers that the library missed.…”