11 July 2025
MediAlbertina is a newly developed AI-based model that enables doctors to quickly and accurately have access to organised medical reports, facilitating better patient screening and the optimisation of healthcare resources.
Traditional manual analysis of medical reports is extremely time consuming but crucial for accurate diagnosis and effective patient screening.
In a recently published scientific study, Luís Elvas - Health Data Engineer in the Breast Cancer Research Programme at the Champalimaud Foundation - and colleagues, have shown that fine-tunning language models can be used for automated analysis of medical reports. This approach can streamline daily clinical workflows, improve categorisation of diseases in patients and offer a tool for new clinical research analysis.
The newly developed model, called MediAlbertina, applies Natural Language Processing (NLP) techniques tailored for the Portuguese healthcare sector, since it was trained on (European Portuguese written) medical reports from Hospital Santa Maria in Lisbon, Portugal.
MediAlbertina automates the analysis of medical reports achieving 96.13% accuracy, enabling fast and reliable diagnosis of conditions, namely aortic stenosis, pneumonia and cancer.
Link to scientific publication: https://www.nature.com/articles/s41598-025-05695-6
