Does BERT Pretrained on Clinical Notes Reveal Sensitive Data?

TL;DR

This study investigates whether BERT models trained on clinical notes can inadvertently reveal sensitive patient information, finding simple methods ineffective but highlighting potential risks with more advanced attacks.

Contribution

The paper introduces probing techniques to assess privacy risks in BERT models trained on clinical data and provides a baseline for future research on data exposure risks.

Findings

Simple probing methods do not extract sensitive info from BERT trained on EHR.

Advanced attack methods may succeed in extracting personal health data.

Experimental setup and baseline models are publicly available for further research.

Abstract

Large Transformers pretrained over clinical notes from Electronic Health Records (EHR) have afforded substantial gains in performance on predictive clinical tasks. The cost of training such models (and the necessity of data access to do so) coupled with their utility motivates parameter sharing, i.e., the release of pretrained models such as ClinicalBERT. While most efforts have used deidentified EHR, many researchers have access to large sets of sensitive, non-deidentified EHR with which they might train a BERT model (or similar). Would it be safe to release the weights of such a model if they did? In this work, we design a battery of approaches intended to recover Personal Health Information (PHI) from a trained BERT. Specifically, we attempt to recover patient names and conditions with which they are associated. We find that simple probing methods are not able to meaningfully extract sensitive information from BERT trained over the MIMIC-III corpus of EHR. However, more sophisticated "attacks" may succeed in doing so: To facilitate such research, we make our experimental setup and baseline probing models available at https://github.com/elehman16/exposing_patient_data_release