A research project at the Technische Hochschule Mittelhessen is concerned with the computer-based coding of medical information in patient files. Health insurance companies, for example, need such data to bill services. The project leaders at THM are the Friedberg professors Dr. Michael Guckert and Dr. Christian Schulze from the Competence Center for Information Technology. You cooperate with Prof. Dr. Mirjam Minor from the Department of Computer Science and Mathematics at Frankfurt Goethe University. Another partner is Minds-Medical from Frankfurt. The company specializes in the development of artificial intelligence for the healthcare industry. The state of Hesse is supporting the project with almost 370,000 euros.
According to the Federal Statistical Office, around 20 million patients are treated in hospitals in Germany every year. For everyone, all medically significant information is documented in patient files. The coding of the data requires highly qualified professionals. The "International Statistical Classification of Diseases and Related Health Problems" published by the World Health Organization alone includes more than 12,000 disease classes in 22 chapters.
Minds-Medical has developed software that automatically encodes patient files and doctor's letters. The company uses machine learning techniques and methods for machine processing of natural language.
The aim of the current project is to use transfer learning methods to make the results usable for all customer groups. This is currently only possible to a limited extent, since the vocabulary used to describe the medical facts differs from clinic to clinic and the data may contain personal information.
The researchers want to make it possible to transfer existing knowledge by using existing medical terminology systems as the basis for classifying patient data. Terms used synonymously in different clinics should be recognized in this way and assigned to the same code. This can significantly simplify the introduction of the coding software in a new clinic.
Potential customers are the almost 2000 hospitals in Germany. According to a study by Minds-Medical, personnel costs for medical coding amount to one billion euros a year. The new technology should also be used by private health insurance companies. There, the software can be used for automated risk analysis for new customers and can identify previous illnesses based on medical reports.
The research project runs for two years and has a total volume of 500,000 euros. It is supported as part of the Hessian "state offensive for the development of scientific and economic excellence" (Loewe).