Artificial intelligence applied to clinical oncology


ModGraProDep is an innovative system that, through artificial intelligence and the design of probabilistic models, improves the prediction of survival indicators in breast cancer patients. This study, published in the journal Artificial Intelligence in Medicine, has been led by Ramon Clèries, professor at the Department of Clinical Sciences of the Faculty of Medicine and Health Sciences of the University of Barcelona (UB) and member of the Institute for Biomedical Research Bellvitge (IDIBELL).

The new work has been developed by a large team of experts in epidemiology, oncology and data management from the Oncology Master Plan-IDIBELL, the University of Barcelona (UB), the Polytechnic University of Catalonia (UPC), the Catalan Institute of Oncology (ICO), the Girona Biomedical Research Institute (IDIBGI), the University of Girona (UdG), the University of Alicante (UA), the CIBER of Epidemiology and Public Health (CIBERESP), the Carlos III Health Institute, the Sant Joan University Hospital of Reus, the Medical Oncology Service of the Josep Trueta Hospital in Girona, the Cancer Registries of Girona and Tarragona, and the entity MC Mutual.


Mathematical modelling: new frontiers in the fight against cancer

Predicting the survival of patients is decisive for evaluating treatments and future evolution scenarios for cancer patients. However, this information often has to be estimated using mathematical models, since there is not a sufficient sample population to calculate these clinical indicators specifically.

This new ModGraProDep methodology (ModelingGraphicalProbabilistcDependencies) has promoted two studies coordinated by Professor Mireia Vilardell, from the Department of Genetics, Microbiology and Statistics at the UB Faculty of Biology, and researcher María Buxó, from IDIBGI. In the first study, ModGraProDep allows simulating a population of cancer patients, from a real database, to identify possible new patterns and calculate indicators (for example, the survival of a patient based on its characteristics). In the second study, ModGraProDep proofs able to predict and assign values probabilistically to variables from which no information could be collected.

The scientific team has also designed a web application that allows obtaining a prediction of survival and mortality risk indicators for each patient, up to a maximum of twenty years.

Scroll to Top