Domènec Puig leads an AI project that accurately predicts the risk of breast cancer recurrence
An international team led by Domènec Puig, professor in the Department of Computer Engineering and Mathematics at the Rovira i Virgili University (URV), has developed an artificial intelligence model capable of predicting with high accuracy the risk of breast cancer recurrence.
The system integrates magnetic resonance images and clinical data from patients to provide a more comprehensive analysis than current methods, as it also takes into account the breast tissue surrounding the tumor. Based on a TabNet neural network, the model operates automatically and can identify subtle patterns-such as symmetry between the breasts or tumor texture-associated with a higher likelihood of recurrence.
In tests with more than 500 patients, the system achieved one of the highest accuracy levels to date and demonstrated high sensitivity in detecting cases at real risk of recurrence. According to Puig, "this sensitivity is crucial to reduce false negatives and ensure proper follow-up for high-risk cases."
The analysis of the results also identified key factors such as irregular tumor texture, lack of symmetry, and the status of hormone receptors, which could become new tools for clinical decision-making.
The model is scalable, interpretable, and applicable in hospitals without the need for invasive or costly genetic tests. The study is part of the European Bosomshield project, within the Marie Skłodowska-Curie Doctoral Networks program, and demonstrates the potential of research led by Domènec Puig to advance toward a more personalized and predictive oncology.
Related elements
Share
