predict the use of existing drugs for new diseases

Researchers from the Polytechnic University of Madrid (UPM) have created drug repositioning method which explains why drug “Y” should be used to treat disease “X”. Previously, machine learning algorithms responsible for predict for which diseases the medicine can be used They could offer no excuses.

Now this new algorithm called “XG4REPO(eXplainable Graphs for Repurencing) presents its results in a clear way and points out the biological mechanisms involved in the prediction. This allows medical experts to test predictions and evaluate the validity of explanations, which in turn leads to stronger forecasts.

Development new drugs are a long and expensive process. For this reason, repositioning of existing drugs has become urgent. The researchers emphasize that this method “has a number of advantages that cannot be ignored.”

This approach significantly reduces development time since you are working with a drug that has already been approved and whose side effects are known. It also eliminates the need to re-run costly security tests. It’s also important to note that many drugs developed in a laboratory fail to reach the market due to their side effects, a problem that can be avoided by repositioning.

On a technical level A. the drug affects a specific biological processHowever, repositioning is about identifying which patterns influenced by a particular drug are evident in other diseases. Here, if two pathologies have a similar pictureand a certain drug is used to treat the former, it is likely that it can be used for the latter.

Researchers say machine learning methods are very good at detecting these patterns, hence the recent focus on repositioning drugs using artificial intelligence. However, artificial intelligence methods are the problem of interpretability.

As a result, a team of researchers from UPM’s ETSI Telecommunications (ETSIT) created “XG4Repo“, a method based on drug reuse through knowledge graphs that foresee illnesses which can be treated with a special compound.

To demonstrate the effectiveness of “XG4REPO”, the researchers tried to predict the use three known drugs in cancer treatment and found that among the algorithm’s predictions, many were already in early clinical trials.

Professor and researcher, Santiago Zazoassured the importance of this promotion: “This mechanism represents another step towards the application of artificial intelligence methods in the medical field.” In turn, Zazo emphasized that this helps them. make analysis easier for experts “from a large amount of data in a short time and speed up the drug preparation process.”