Assessing the Reidentification Risks Posed by Deep Learning Algorithms Applied to ECG Data

Abstract

ECG (Electrocardiogram) data analysis is one of the most widely used and important tools in cardiology diagnostics. In recent years the development of advanced deep learning techniques and GPU hardware have made it possible to train neural network models that attain exceptionally high levels of accuracy in complex tasks such as heart disease diagnoses and treatments. We investigate the use of ECGs as biometrics in human identification systems by implementing state-of-the-art deep learning models. We train convolutional neural network models on approximately 81k patients from the US, Germany and China. Currently, this is the largest research project on ECG identification. Our models achieved an overall accuracy of 95.69%. Furthermore, we assessed the accuracy of our ECG identification model for distinct groups of patients with particular heart conditions and combinations of such conditions. For example, we observed that the identification accuracy was the highest (99.7%) for patients with both ST changes and supraventricular tachycardia. We also found that the identification rate was the lowest for patients diagnosed with both atrial fibrillation and complete right bundle branch block (49%). We discuss the implications of these findings regarding the reidentification risks of patients based on ECG data and how seemingly anonymized ECG datasets can cause privacy concerns for the patients.

Review of Ex Vivo Cardiac Electrical Mapping and Intelligent Labeling of Atrial Fibrillation Substrates

“…In addition, because of the individual variability of ECG signals, the apparently anonymous ECG data may be a threat to the personal privacy of users [63] .…”

MBFRDH: Design of a Multimodal Bioinspired Feature Representation Deep Learning Model for Identification of Heart-Diseases

“…This has been the case for some of the ARRs in the heart. Signals from electrocardiograms (ECGs) were put through a series of tests, which are reported in [30,31,32,33], and the findings demonstrated the existence of both deterministic chaos and nonlinear dynamics. The chaotic ventricular response that is seen in this case was corroborated by the findings of an investigation that used a nonlinear prediction approach and analyzed ECG recordings made during atrial fibrillation for their predictability and sensitivity to starting conditions.…”

“…This is part of a novel strategy that has been developed. In the study referenced as [30], fragmentation indicators were employed in both classic linear and nonlinear heart rate variability (HRV) studies in order to improve classification performance of cardiac disorders. The vast majority of feature-based machine learning algorithms depend on HRV analysis to provide highquality ECG diagnostic results; however, there is no assurance that these results will be robust.…”