Artificial intelligence learns to determine human reactions to drugs

U.S. scientists have created an artificial intelligence model that could significantly improve accuracy, reduce the time and cost of the drug development process. An article by researchers from the City University of New York (CUNY Graduate Center) was published in the journal Nature Machine Intelligence.

Usually, the process of creating drugs requires several years: first, scientists must find the right compound, test it, and then get approval. A new neural network called CODE-AE can test new drug compounds and accurately predict their effectiveness in humans. Trials have theoretically been able to find personalized drugs for more than 9,000 specific patients.

"Using CODE-AE, we tested 59 drugs for 9,808 patients with cancer. Our results are consistent with existing clinical observations," the researchers wrote in their paper.

The work analyzed data from patient cell lines. An array of real cell state parameters was loaded into the system and it analyzed how they worked. Then a drug was "added" to it and then it showed how the original virtual cells reacted to this or that composition.

Accurate and reliable prediction of a patient's response to a new chemical compound is crucial to the discovery of new drugs and the selection of a drug for a particular patient from existing ones. However, it is not possible to perform early testing directly on humans. This is a major factor in the high cost and low productivity of drug development.

"Our new machine learning model can solve the problem. CODE-AE takes advantage of biology-inspired design and the benefits of several recent advances in machine learning," said Lei Xie, professor of computer science, biology and biochemistry and senior author of the paper.

The research team's next challenge is to develop a way to reliably predict the effects of the new drug's concentration and metabolism on the human body. The researchers also noted that the neural network could be tuned to accurately predict the side effects of drugs in a particular person.

Follow NEWS.am Medicine on Facebook and Twitter