Fig. 2: The mutual link between the data modalities, generative models, and research objectives found by the research objectives.
From: A review on generative AI models for synthetic medical text, time series, and longitudinal data
This Figure shows an overview of the findings in the paper. In summary, generative adversarial networks (GANs) were dominantly employed for generating medical time series. Large language models (LLMs) have been widely used for generating synthetic texts. The variational auto-encoder (VAE) method was employed in a minority of the studies, equally for generating longitudinal and text data, but it was not used for time series. Probabilistic models were dominantly utilized for longitudinal data.
