Two researchers at Fundação Getulio Vargas’ School of Applied Mathematics (FGV EMAp), Alberto Paccanaro and Diego Galeano, have developed a new method for predicting the frequency of medication side effects. Together with Dr. Shantao Li of Stanford University and Professor Mark Gerstein of Yale University, they carried out a study on this subject that was published recently in Nature Communications.

Identifying the frequency of medication side effects among the public is a key challenge for the process of developing new drugs. Until now, this frequency could only be determined experimentally, through clinical trials, in a process with many limitations. Some side effects are only discovered after the drug is launched on the market.

The method developed by Paccanaro and Galeano is therefore a breakthrough. It is the first computational method for predicting the frequency of medication side effects in the population. They developed an artificial intelligence algorithm based on matrix decomposition, which learns about a small group of components (latent factors) that encode biological interactions between drugs and generate side effects involving multiple types of drugs.