Artificial Intelligence for Drug Repositioning

We at Kunumi, together with the Artificial Intelligence Laboratory (LIA) of the Federal University of Minas Gerais (UFMG), develop artificial intelligence models to discover new uses for known drugs.

The process, known as drug repositioning, consists of finding a new treatment opportunity with a drug that is already marketed or that has failed tests for its initial goals. Sildenafil citrate (Viagra) is a famous example of this process: originally created to treat hypertension, it has been repositioned to treat erectile dysfunction. Repositioned drugs account for approximately 30% of US Food and Drug Administration (FDA)-approved drugs in recent years. A chemical substance is given for the treatment of a disease because it demonstrates a particular mechanism of action that affects the biological phenomena related to that disease.

The process of developing and approving a new drug is expensive and long, taking between 10 and 17 years![1] The National Health Surveillance Agency (ANVISA), the body responsible for evaluating and approving a new drug in Brazil, submits the drug to several tests before its commercialization to guarantee its effectiveness and safety. Tested on animals and finally on people, known as the clinical phase, the drug needs to be approved in all tests before it can be marketed. This is the great advantage of drug repositioning: When a new possibility of use for a drug is identified, it can skip the initial phases and go straight to the testing phase in people, reducing time and expenses. That is why it is so important to search for new techniques and strategies in an attempt to find new drugs or to optimize existing techniques.

Through machine learning techniques, more precisely supervised learning in graphs, we analyze databases on drugs, diseases and proteins and create models capable of predicting new uses for drugs.

The researchers also searched the biomedical literature to validate the results, verifying that what is suggested by the model is confirmed by the biomedical literature.

Some repositioning candidates suggested by the model were found, which were also mentioned in the biomedical literature.

The work was recently accepted for publication at the “International Joint Conference on Neural Networks (IJCNN)” which will take place in July 2020.

Other articles: