A Scoping Review of Synthetic Data Generation by Language Models in Biomedical Research and Application: Data Utility and Quality Perspectives.

A notable trend is the model selection between open-source and closed-source LLMs, such as Llama vs GPT-4.

also in 2,135 other papers

There is a clear trend towards generating multimodal synthetic datasets to simulate more real-world and complex clinical scenarios, such as radiology [ 7 , 70 ], pathology [ 72 ], and emergency care [ 41 ].

also in 12,254 other papers