作者：Li Jin, Cairns Benjamin J., Li Jingsong；等
来源：NPJ DIGITAL MEDICINE，卷:6 期:1 页(号):98 出版时间:MAY 2023
Electronic health records (EHRs) offer opportunities for AI research, but patient privacy concerns limit data sharing. This study introduces EHR-M-GAN, a generative adversarial network that synthesizes mixed-type timeseries EHR data mimicking clinical decision-making. Experimental evaluations demonstrate that EHR-M-GAN outperformed benchmarks in generating high-fidelity, privacy-preserving clinical timeseries, improving predictive models for patient outcomes when augmenting training data.
