Swedish Medical Language Data Lab

Purpose

Natural Language Processing (NLP) has the potential to provide insights from the large amounts of medical text data that are generated daily in the Swedish healthcare and life sciences sector. Patterns and concepts that today seem to be hidden in this unstructured data could be brought to light by applying medical NLP. By searching, analyzing, and interpreting the data sets, these insights can give us a deeper understanding of healthcare quality, improving practices, and effecting better results for patients.

The Swedish Medical Language Data Lab aims to develop medical language models and to increase the accessibility to medical data. This is handled in accordance with the ethical, legal, and technical processes that are required to manage sensitive medical data in text form. The models are being developed for applications within healthcare, dental care, and data-driven research and innovation within the life sciences sector.

The initial use cases in the project are provided by the stakeholders and data owners in the project. The guiding criteria for the project are based on the potential benefit, the end-users, and personal integrity. 

The establishment of a national knowledge node in the field of NLP began with the Swedish Language Data Lab. Models, methods, and findings specific to the medical domain will be shared during the course of this project.

Project Team

The project is coordinated by Sahlgrenska Science Park. Peltarion, and RISE (Research Institutes of Sweden). AI Sweden is the expert resource for the project. The stakeholders and data owners are Region Halland, Folktandvården Västra Götalandsregionen, and Sahlgrenska University Hospital.

The initial interest group consists of Sahlgrenska Academy at the University of Gothenburg, Essity Hygiene and Health AB, and Språkbanken at the University of Gothenburg. A reference group consisting of organizations from the partners’ networks will also be set up.

AI Sweden is responsible for mapping the procedures of making data and models available and for the distribution of the results.

Results 

NLP in the Swedish Medical Language Data Lab 
Here you can read a description of the technology NLP,  and a summary of findings, models, and methods used for NLP analysis of medical health records in Region Halland and Folktandvården Västra Götaland. Produced by RISE and Peltarion. 

Techlaw NLP article and Techlaw webinar pictures
Overview of selected legal aspects concerning personal data when training NLP models with patient record texts. The purpose of this article is to shed light on the use of anonymized patient records for training NLP models at a public healthcare provider.
Financed by Region Halland and produced by the law firm TechLaw in collaboration with AI Sweden. (in Swedish) 

Contact