A scientist from the University of Navarra investigates in Canada how to use AI to advance the fight against cancer.
Rubén Armañanzas collaborates with a leading center in Toronto to detect common genetic patterns in different types of tumors.
Detecting common genetic patterns in different types of cancer can improve their diagnosis and facilitate more personalized treatments. This is the line of research being carried out by Rubén Armañanzas, manager of the Digital Medicine Lab at Institute for data Science and Artificial Intelligence (DATAI) of the of the University of Navarra, during his stay at the Princess Margaret Cancer Centre in Toronto (Canada), one of the most prestigious cancer hospitals in the world.
For three months, Armañanzas has been collaborating with the BHKLab, a team led by Dr. Benjamin Haibe-Kains, an expert in artificial intelligence applied to cancer. Together they are working on the development of algorithms capable of analyzing large volumes of genetic data in search of genomic signatures, i.e. common genetic patterns in tumors originating in different organs.
This approach, known as pan-cancer analysis, studies the genetic similarities between several types of cancer that are normally investigated separately. This subject analysis can open new avenues to better understand the disease and design more precise treatments.
"There are still few genetic signatures that work well across different types of cancer. We want to develop an algorithm that detects genetic patterns that repeat consistently and reliably across multiple tumors. This could be a core topic for moving towards more precise and effective medicine," explains the researcher.
Pan-cancer analysis is gaining relevance worldwide because it proposes a change in the way tumors are studied: they are no longer analyzed only by their location, but also by their genetic characteristics. This broader vision makes it possible to find more cross-cutting solutions to a complex disease such as cancer.
Mathematical reliability applied to cancer
The methodology used in this work is based on conformal prediction, an advanced statistical technique that in this study allows us to reliably identify sets of core topic genes with similar behavior in different tumors. This methodology has been perfected by the Amañanzas team in recent years.
"In this project we are using data from 21 immunotherapy clinical trials with 10 different types of cancer such as melanoma, lung or bladder, among others. Thanks to this partnership, we are working with a very broad and diverse clinical database , which gives us greater strength to validate our models," adds the researcher.
In addition to the scientific impact, the stay strengthens the partnership between international research teams and opens the door to new opportunities for students and researchers of the University of Navarra.
The project is funded by the Ministry of Science, Innovation and Universities, within the program of excellence stays for professors and researchers in high-level foreign centers.
