Medicine is continually making progress. For the majority of cases we know, there is a solution. One of the most prominent problems is still a lack of time. Especially medical doctors suffer from it. They cannot afford to take the time they need to investigate rare medical cases. Medical Named Entity Recognition and Named Entity Linking approaches are supposed to process the findings while medical professionals type theire reports. Found entities are used to search for relevant publications or even patients with similar symptoms. The project TastyOnFhir aims to analyze electronic health records. At first, we recognize medical entities in patients' data. Secondly, we use the Unified Medical Language System (UMLS) as a knowledge base and link every found medical concept to its unique identifier. Based on these identifiers our system can build Fast Healthcare Interoperability Resources (FHIR) packages that can be interpreted by third-party software applications.
An overview of the results can be found here: pdf1, pdf2, pdf3.
Interactive Named Entity Recognition
The goal of the TraiNER (Train Named Entity Recognition) project was to develop a system that illustrates an Active Learning Process in the field of NER. For this purpose, a system consisting of several components was implemented.
It allows the user to train a NER model by marking spans of text in an textmarker-like fashion. In SS 2019 TraiNER has been further developed.
An overview of the results can be found in the presentation.
The results from WS 2017/2018 can be seen here.