Artificial Intelligence Predicts Side Effects Of Drugs Combination
Advances in artificial intelligence are gradually helping researchers predict the side effects of drug combinations. This information will help to improve patient safety.
People who need medication rarely take only one class of drugs. Many people who take drugs for health reasons take more than five drugs a day. According to the US Centers for Disease Control and Prevention, in the past 30 days, the United States has used at least. The proportion of people taking a drug is 48.9%; in the past 30 days, the proportion of people using three or more drugs is 23.1%; and the proportion of using five or more drugs in the past 30 days is 11.9%. US data will be consistent with data from many other high-income countries.
From ubiquitous aspirin to the most sophisticated drugs on the market, all drugs have side effects, many of which are mild, while others are more severe. For example, any effective drug can cause nausea or stomach upset. Others can cause allergies such as rashes or dry mouth. More serious internal bleeding may occur.
Although the side effects of a certain drug are well known, the dilemma faced by doctors is that they know that these drugs have side effects with each other, often because they are too complicated to predict, and there are few clinical trials in this area. This situation is unlikely to change in this area, as it is unrealistic to apply this test to a combination of drugs. To understand its complexity, there are approximately 5,000 licensed drugs with 1000 side effects known. This means that once mixed with different drugs, there will be 125 billion side effects.
However, the solution is at hand. Computer scientists at Stanford University have developed how to use artificial intelligence to predict the side effects of a combination drug. Known as the Decagon, this new system helps doctors make more informed decisions about which drug to use. The platform also has the potential to provide researchers with a way to find new drug combinations to help treat complex diseases.
Decagon trains through deep learning and conducts platform testing on known drug combinations. Currently, artificial intelligence can only evaluate a pair of drugs; the future goal is to use the system for a combination of three or more drugs. The study was recently submitted to the Chicago International Computational Biology Society in July 2018.