Generating Multi-label Discrete Patient Records using Generative Adversarial Networks

Abstract

Access to electronic health record (EHR) data has motivated computational advances in medical research. However, various concerns, particularly over privacy, can limit access to and collaborative use of EHR data. Sharing synthetic EHR data could mitigate risk. In this paper, we propose a new approach, medical Generative Adversarial Network (medGAN), to generate realistic synthetic patient records. Based on input real patient records, medGAN can generate high-dimensional discrete variables (e.g., binary and count features) via a combination of an autoencoder and generative adversarial networks. We also propose minibatch averaging to efficiently avoid mode collapse, and increase the learning efficiency with batch normalization and shortcut connections. To demonstrate feasibility, we showed that medGAN generates synthetic patient records that achieve comparable performance to real data on many experiments including distribution statistics, predictive modeling tasks and a medical expert review. We also empirically observe a limited privacy risk in both identity and attribute disclosure using medGAN.

Accepted at Machine Learning in Health Care (MLHC) 2017

References in corpus (7)

• Conditional Generative Adversarial Nets
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• NIPS 2016 Tutorial: Generative Adversarial Networks
• The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
• GANS for Sequences of Discrete Elements with the Gumbel-softmax Distribution
• Mixed-Variate Restricted Boltzmann Machines
• Modeling documents with Generative Adversarial Networks

Cited by in corpus (72)

• Data Synthesis based on Generative Adversarial Networks
• Differentially Private Generative Adversarial Network
• GAN-Leaks: A Taxonomy of Membership Inference Attacks against Generative Models
• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Deep learning for comprehensive forecasting of Alzheimer's Disease progression
• Synthesizing Tabular Data using Generative Adversarial Networks
• Using GANs for Sharing Networked Time Series Data: Challenges, Initial Promise, and Open Questions
• NetGAN: Generating Graphs via Random Walks
• A Survey on Data Collection for Machine Learning: a Big Data -- AI Integration Perspective
• Road Segmentation for Remote Sensing Images using Adversarial Spatial Pyramid Networks
• Differentially Private Synthetic Medical Data Generation using Convolutional GANs
• FedDPGAN: Federated Differentially Private Generative Adversarial Networks Framework for the Detection of COVID-19 Pneumonia
• A Multifaceted Benchmarking of Synthetic Electronic Health Record Generation Models
• When Machine Learning Meets Privacy: A Survey and Outlook
• Modeling Tabular data using Conditional GAN
• Adversarial Attacks Against Deep Generative Models on Data: A Survey
• Natural Language Generation for Electronic Health Records
• PPGAN: Privacy-preserving Generative Adversarial Network
• Leveraging Generative AI Models for Synthetic Data Generation in Healthcare: Balancing Research and Privacy
• CTAB-GAN: Effective Table Data Synthesizing
• Synthesize High-dimensional Longitudinal Electronic Health Records via Hierarchical Autoregressive Language Model
• Generative adversarial networks in time series: A survey and taxonomy
• DermGAN: Synthetic Generation of Clinical Skin Images with Pathology
• Generating Multi-Categorical Samples with Generative Adversarial Networks
• FairGAN: Fairness-aware Generative Adversarial Networks
• Copula Flows for Synthetic Data Generation
• Generative Adversarial Networks for Electronic Health Records: A Framework for Exploring and Evaluating Methods for Predicting Drug-Induced Laboratory Test Trajectories
• Synthesis of Realistic ECG using Generative Adversarial Networks
• Two Directions for Clinical Data Generation with Large Language Models: Data-to-Label and Label-to-Data
• Generative Adversarial Networks: A Survey Towards Private and Secure Applications
• Synthesizing time-series wound prognosis factors from electronic medical records using generative adversarial networks
• Improving Missing Data Imputation with Deep Generative Models
• CorGAN: Correlation-Capturing Convolutional Generative Adversarial Networks for Generating Synthetic Healthcare Records
• Measuring the quality of Synthetic data for use in competitions
• Adversarial Training for Disease Prediction from Electronic Health Records with Missing Data
• Synthetic Observational Health Data with GANs: from slow adoption to a boom in medical research and ultimately digital twins?
• Teacher-Student Compression with Generative Adversarial Networks
• GAN-based Tabular Data Generator for Constructing Synopsis in Approximate Query Processing: Challenges and Solutions
• Categorical EHR Imputation with Generative Adversarial Nets
• Generating Electronic Health Records with Multiple Data Types and Constraints
• Synthetic Data -- Anonymisation Groundhog Day
• Tubular Shape Aware Data Generation for Semantic Segmentation in Medical Imaging
• Generation of Differentially Private Heterogeneous Electronic Health Records
• Generative Synthesis of Insurance Datasets
• DAG-WGAN: Causal Structure Learning With Wasserstein Generative Adversarial Networks
• Imagining an Engineer: On GAN-Based Data Augmentation Perpetuating Biases
• Causal-TGAN: Generating Tabular Data Using Causal Generative Adversarial Networks
• Causal Adversarial Network for Learning Conditional and Interventional Distributions
• Generation of Synthetic Electronic Health Records Using a Federated GAN
• Data Augmentation for Enhancing EEG-based Emotion Recognition with Deep Generative Models
• Competitive Multi-Agent Deep Reinforcement Learning with Counterfactual Thinking
• Effective and Privacy preserving Tabular Data Synthesizing
• Holdout-Based Fidelity and Privacy Assessment of Mixed-Type Synthetic Data
• Generating Synthetic Text Data to Evaluate Causal Inference Methods
• PrivGen: Preserving Privacy of Sequences Through Data Generation
• Ward2ICU: A Vital Signs Dataset of Inpatients from the General Ward
• Synthesising Multi-Modal Minority Samples for Tabular Data
• EMIXER: End-to-end Multimodal X-ray Generation via Self-supervision
• Composable Generative Models
• ODVICE: An Ontology-Driven Visual Analytic Tool for Interactive Cohort Extraction
• A Differentially Private Multi-Output Deep Generative Networks Approach For Activity Diary Synthesis
• OCT-GAN: Neural ODE-based Conditional Tabular GANs
• Scalable Bayesian Nonparametric Clustering and Classification
• G2R Bound: A Generalization Bound for Supervised Learning from GAN-Synthetic Data
• Adversarially-learned Inference via an Ensemble of Discrete Undirected Graphical Models
• Multiple Imputation for Biomedical Data using Monte Carlo Dropout Autoencoders
• Training face verification models from generated face identity data
• Minority Class Oversampling for Tabular Data with Deep Generative Models
• Sequence Preserving Network Traffic Generation
• Towards Generating Real-World Time Series Data
• Improving Tuberculosis (TB) Prediction using Synthetically Generated Computed Tomography (CT) Images
• Causality Learning With Wasserstein Generative Adversarial Networks