%0 Journal Article
%T TimEHR: Image-Based Time Series Generation for Electronic Health Records
%V 29
%N 12
%P 9170-9180
%W https://infoscience.epfl.ch/entities/publication/0b042ae8-fd8c-40fd-8b29-32a70b82172d
%U https://ieeexplore.ieee.org/document/11027528
%X Time series in Electronic Health Records (EHRs) present unique challenges for generative models, such as irregular sampling, missing values, and high dimensionality. In this paper, we propose a novel generative adversarial network (GAN) model, TimEHR, to generate time series data from EHRs. In particular, TimEHR treats time series as images by using 2D convolutional kernels and is based on two conditional GANs. The first GAN generates missingness patterns, and the second GAN generates time series values based on the missingness pattern. Experimental results on three real-world EHR datasets show that TimEHR outperforms state-of-the-art methods in terms of fidelity, utility, and privacy metrics.
%J IEEE Journal of Biomedical and Health Informatics
%A Karami, Hojjat
%A Hartley, Mary-Anne
%A Atienza, David
%A Ionescu, Anisoara
%D 2025-12
%K Convolutional neural nets
Convolutional neural networks
Data privacy
Electronic health records (EHRs)
Electronic medical records
Generative adversarial networks
Irregularly sampled time series
Mathematical models
Metadata
Predictive models
Time series analysis
Time series generation
scientific-publication