Using Artificial Intelligence to Generate Synthetic Health Data

Résumé du livre

Generating synthetic data enables making sensitive data sets available to the research community. This report utilizes two off-the-shelf methods to generate synthetic health data. One method, synthpop, is based on standard statistical techniques. The other, CTGAN, is a deep learning generative adversarial network. The authors compare the performance of the methods and discuss the reasons for which synthpop outperforms CTGAN on these data.