In the public project FHIR-Starter, Fraunhofer IESE and its project partners are researching ways of automatically analyzing full-text medical documents, such as findings, using large language models and making them available as FHIR resources in an interoperable manner - here we place particular emphasis on safeguarding and avoiding hallucinations.
Last modified:
In the FHIR-Starter project, we are developing a software service that uses Large Language Models (LLM) and Natural Language Processing (NLP) to analyze full-text documents and convert them into the FHIR standard medical data format. The service offers open interfaces for service providers, healthcare software providers and secondary users of healthcare data to enable structured integration of the data into their systems. This improves the interoperability and accessibility of the data, which brings direct benefits for medical care, as staff are relieved and specialist personnel are provided with information more easily. It also promotes the structured collection of medical data for researchers and increases Germany's attractiveness for medical research.
Fraunhofer Institute for Experimental Software Engineering IESE