AI system can predict side effects of drug combinations

Stanford researchers designed an artificial intelligence system that can predict potential side effects for two-drug combinations, and plan to expand it for patients who take three or more medications simultaneously.

By Allen Cone
AI system can predict side effects of drug combinations
An artificial intelligence system was shown to successfully predict the potential side effects from the combination of two drugs, researchers at Stanford report. Photo by Brian Kersey/UPI | License Photo

July 10 (UPI) -- Researchers have developed an artificial intelligence system to predict potential side effects from two-drug combinations.

With the Decagon system, researchers at Stanford University believe that can doctors make better decisions about combinations of drugs to prescribe, but the right combo can be determined to treat complex diseases. The findings were presented this week at the 2018 meeting of the International Society for Computational Biology in Chicago and published in the journal Bioinformatics.


The Centers for Disease Control and Prevention estimates that 23 percent of Americans take two or more prescription drugs, and 39 percent over age 65 take five tor more.

"It's practically impossible to test a new drug in combination with all other drugs, because just for one drug that would be five thousand new experiments," said Dr. Marinka Zitnik, a postdoctoral fellow in computer science at Stanford, said in a press release. "Truly we don't know what will happen."

RELATED Patients, advocates raise concern about opioid guidance at FDA hearing

The researchers said there are about 1,000 known side effects of drugs on the market -- creating nearly 125 billion possible side effects between all possible pairs of drugs.


Fifty-three percent of drug side effects are known to occur in less then 3 percent of the documented drug combinations.

The researchers determined how the more than 19,000 proteins in our bodies interact with each other and how different drugs affect these proteins. They then designed a way to identify patterns in how side effects arise based on how drugs target different proteins in more than 4 million known associations.

RELATED Combining antibiotics alters effectiveness, study says

They accomplished this task with deep learning, a form of artificial intelligence modeled after the brain.

To test their system, they looked to see if its predictions came true.

Searching medical literature for evidence of side effects predicted by Decagon, they found that five out of the 10 have recently been confirmed.

RELATED Model finds drug combos to treat triple-negative breast cancer

For example, Decagon found the combination of atorvastatin, a cholesterol drug, and amlopidine, a blood pressure medication, could lead to muscle inflammation. And it was right -- a 2017 case report suggested the drug combination had led to a dangerous kind of muscle inflammation.

The researchers found a median of 159 side effects per drug, with the most common side effects being nausea, vomiting, headache, diarrhea and dermatitis.

"It was surprising that protein interaction networks reveal so much about drug side effects," said Dr. Jure Leskovec, an associate professor of computer science at Stanford.


The researchers hope to extend their system to more than two-drug combos, as well as create a more user-friendly tool for doctors.

"Today, drug side effects are discovered essentially by accident and our approach has the potential to lead to more effective and safer health care." Leskovec said.

Latest Headlines


Follow Us