Supporting physicians in making complex and rare clinical decisions is the goal of MedLink, a tool developed by researchers at INESC TEC, which won the Best Demo Paper award at the European Conference on Information Retrieval—one of the most prestigious conferences in Europe in the field of information retrieval.
Making medical decisions often relies solely on the healthcare professional’s knowledge. In more complex and rare cases, it may also involve peer opinions and support from clinical reports with similar symptoms and indicators. However, accessing these reports can be challenging due to the wide range of platforms used to publish and catalogue available documentation. To simplify access, INESC TEC researchers developed MedLink, a free-text tool that retrieves and ranks clinical reports published in conferences and journals.
The scientific relevance of the proposed solution was recognized with the Best Demo Paper award at the European Conference on Information Retrieval, a prestigious European event in the field of information retrieval with a core rank of A.
By training two models derived from BERTimbau—a Portuguese-language variant of BERT—the researchers developed the component of the tool dedicated to sentence similarity: one model is responsible for selecting articles, and the other ranks the proposed selection. The data used included published scientific articles (specifically clinical case reports) from the Portuguese Society of Internal Medicine journal. The tool’s development also included the collaboration of physician Alexandra Mendes from the Braga Local Health unit during the model evaluation phase.
In practical terms, MedLink “was designed so that, when entering a medical report, healthcare professionals receive a ranked list of similar clinical cases already published,” explains Nuno Guimarães, one of the INESC TEC researchers involved in developing the tool, alongside Luís Filipe Cunha, Ricardo Campos, and Alípio Jorge. He also confirms that the solution “can be adapted to integrate clinical data from hospitals themselves or the National Health System, thereby significantly expanding its potential application and impact.”
The recognition at the European Conference on Information Retrieval is, for Nuno Guimarães, “a significant acknowledgment of the innovation and practical applicability of the presented demo, showcasing not only its scientific relevance but also its potential real-world impact in clinical settings.” This is the second consecutive year the recognition has been awarded, following the same group of INESC TEC researchers winning the same prize in 2024 with the Physio solution.
The INESC TEC researchers mentioned in the article are affiliated with INESC TEC, the Faculty of Sciences of the University of Porto, UBI, the University of Minho, and Ci2 – Smart Cities Research Center.
News, current topics, curiosities and so much more about INESC TEC and its community!
