A review of Generative Adversarial Networks for Electronic Health Records: applications, evaluation measures and data sources

Abstract

Electronic Health Records (EHRs) are a valuable asset to facilitate clinical research and point of care applications; however, many challenges such as data privacy concerns impede its optimal utilization. Generative Adversarial Networks (GANs) show great promise in generating synthetic EHR data by learning underlying data distributions while achieving excellent performance and addressing these challenges. This work aims to review the major developments in various applications of GANs for EHRs and provides an overview of the proposed methodologies. For this purpose, we combine perspectives from healthcare applications and machine learning techniques in terms of source datasets and the fidelity and privacy evaluation of the generated synthetic datasets. We also compile a list of the metrics and datasets used by the reviewed works, which can be utilized as benchmarks for future research in the field. We conclude by discussing challenges in GANs for EHRs development and proposing recommended practices. We hope that this work motivates novel research development directions in the intersection of healthcare and machine learning.

Outcome Prediction in Critically-Ill Patients with Venous Thromboembolism and/or Cancer Using Machine Learning Algorithms: External Validation and Comparison with Scoring Systems

“…However, the accuracy appeared to remain constant despite the adoption of the SMOTE oversampling technique. Since SMOTE is a deterministic resampling method that selects examples being close in the local feature space, a probabilistic approach, such as Generative Adversarial Networks [52], which originally have been used in the area of image processing to produce synthetic but realistic images, could be used to produce new samples within a medical data analysis framework by learning from the overall class distribution. Moreover, the experimentation and extraction of the best-performing ML model in the custom approach is time-consuming since it requires substantial human and computational effort, artificial intelligence expertise, and extensive tuning of hyper-parameters; for this reason, automated ML tools are becoming popular among non-specialists in this area.…”

“…Shapley additive explanation (SHAP) analysis could be used to explain the output of our predictive model [65]. Handling of the high imbalance ratio of the datasets could be performed with other advanced resampling methods, such as Generative Adversarial Networks [52].…”

Generating Synthetic Electronic Health Record Data Using Generative Adversarial Networks: Tutorial

“…Below, we provide a brief overview of these metrics. For more comprehensive details, we point readers to several recent publications in the field [ 18 , 19 , 21 ], which provide in-depth explanations of how these metrics are designed.…”

Artificial Intelligence in Medicine: Text Mining of Health Care Workers’ Opinions

“…Data sets from a given setting are used by a local server and are not shared with other entities; however, the algorithm being developed takes advantage of all the data sets in the network of servers [45][46][47]. Another approach for minimizing data privacy issues is the use of generative adversarial networks to generate data sets that replicate the statistical distributions of the original data sets and using these generated data sets as training sets for developing the algorithms [48,49].…”