AI-generated patient records available for innovation research

The data from patient records contain valuable information for the development of healthcare innovations but are privacy-sensitive. Synthetic electronic patient records created using generative AI provide a practical alternative.

Synthetic data are artificially generated data that mimic the statistical characteristics, patterns, and structures of real data but do not contain actual observations of individuals or entities. A team of researchers from, among others, Hogeschool Rotterdam and the Medical Delta Living Lab 'Data supported healthcare & innovation'  developed a methodology to use AI-generated patient records for innovation research.

Developing Innovative Healthcare Applications Faster

The methodology generates electronic patient records using a few real examples from everyday practice. The structure, language, and informational content of existing patient records serve as input, but they are free from identifiable personal data. This opens the door to safe research and education without legal or ethical barriers. Using this method, researchers can develop innovative healthcare applications more quickly, without waiting for lengthy approval procedures.

The results have been published in the international journal Frontiers in Artificial Intelligence. The full methodology, including code and example data, is available open source on GitHub, allowing anyone to work with it (Click here).

Read more:

Hogeschool Rotterdam publishes first: privacy-friendly synthetic healthcare data with AI – Hogeschool Rotterdam