Digi­ta­liza­tion in health­care through AI-supported software

Together with the Fraun­hofer Institute and the Berlin Institute for Health Research at Charité, AI pioneer Insiders is deve­lo­ping a solution for struc­tu­ring medical data in the “FHIR Starter” project.

Insiders is par­ti­ci­pa­ting as a key tech­no­logy partner in the FHIR Starter research project. The project is being led by the Fraun­hofer Institute for Expe­ri­mental Software Engi­nee­ring IESE and imple­mented in col­la­bo­ra­tion with the Berlin Institute for Health Research at Charité.

Until now, medical infor­ma­tion has often been exchanged via tra­di­tional doctor’s letters in PDF format. The manual transfer of relevant data into practice manage­ment systems (PMS) or hospital infor­ma­tion systems (HIS) is time-consuming and prone to errors. The aim of the “FHIR-Starter” project is to develop an inno­va­tive software service that uses arti­fi­cial
By using large language models (LLMs) and natural language pro­ces­sing (NLP), unstruc­tured full-text documents are to be analyzed and converted into stan­dar­dized data struc­tures. This enables seamless inte­gra­tion into existing IT systems in hospitals and practices, repre­sen­ting a decisive step toward digi­tiza­tion and auto­ma­tion in the health­care sector.

Key requi­re­ments for the solution are to guarantee the quality and security of the struc­tured data and to ensure GDPR-compliant pro­ces­sing of sensitive medical data.

FHIR-Starter is funded by the Federal Ministry for Economic Affairs and Energy with €1.64 million as part of the inno­va­tion com­pe­ti­tion “Gene­ra­tive AI for SMEs.”

Insiders takes on deve­lo­p­ment of large language model

As an expert in AI-supported document pro­ces­sing, Insiders is par­ti­cu­larly involved in the deve­lo­p­ment of the large language model (LLM), which, as a core component of the software, auto­ma­ti­cally extracts relevant infor­ma­tion from medical documents such as doctor’s letters. This struc­tured infor­ma­tion is then converted into the stan­dar­dized FHIR format,
which was developed spe­ci­fi­cally for the exchange of health data and gives the project its name.

“We are lever­aging our many years of expertise in AI-supported data extra­c­tion to automate the pro­ces­sing of medical documents,” explains Christian Schön, Product Owner OvAition at Insiders. „By deve­lo­ping a powerful LLM, we are enabling medical data to be effi­ci­ently struc­tured and made usable. The auto­ma­tion of admi­nis­tra­tive tasks reduces the workload on
medical pro­fes­sio­nals and minimizes errors. This allows doctors to con­cen­trate on their core task again – providing the best possible care for their patients.“

In addition to deve­lo­ping the LLM, Insiders is also involved in other key project tasks, including iden­ti­fying relevant data, deve­lo­ping a prototype, and publi­shing sci­en­tific results at spe­cia­list con­fe­rences.

Versatile appli­ca­tions for a digital future

The struc­tured pro­ces­sing of medical data opens up a wide range of pos­si­bi­li­ties: Hospitals and medical practices can optimize their processes in findings manage­ment and patient docu­men­ta­tion, medical research insti­tutes gain easier access to relevant infor­ma­tion, and patients benefit from more efficient and better care.

Within the three-year project, which started in February 2025, Insiders is making an important con­tri­bu­tion to the digi­tiza­tion of health­care in Germany.