US scientists use mathematical models to predict drug side effects

After the introduction of a drug, it may take several years for the doctor to find serious, possibly fatal, side effects that did not appear in the trial. If you can predict before the side effects appear to be much better.

A team of scientists from Boston Children's Hospital designed a mathematical prediction model to achieve this goal.

Their research results are published in the online edition of the journal Science Translation Medicine.

This is a significant improvement over the expensive, inefficient, and slow systems that are currently widely used to test drug safety.

Alan Goldberg of the Johns Hopkins-Bloomberg School of Public Health, which was not involved in the study, said: "The development of new drugs requires the use of many different biological systems and requires the use of two types of animals. Its safety and efficacy are tested, usually in rodents and dogs, and the rate of drug failure from an animal test to human clinical trials exceeds 90%. Drugs that pass human clinical trials will enter the market and will be used on a large number of patients. In the first phase, there is still a high chance of unexpected adverse reactions."

Scientists at Boston Children’s Hospital used a mathematical method called a pharmacological network model. They collected 2005 data from several websites and commercial websites Lexicom p. Lexicom p collects information that the federal government requires to insert in drug packaging.

Kami, head of the research team, said that the data includes everything people know about a certain drug: chemistry, type, and known safety issues.

They used this model to calculate the response of 809 drugs and 852 adverse drugs, and compared the results with those of the drugs that people discovered later.

Kami said that the model accurately predicted the 42% of adverse drug reactions that appeared later. It predicts which of the adverse drug reactions will not occur with a precision of 95%.

Kami said that if this approach is adopted, safety experts may be able to test their potential adverse effects before drug distribution.

Goldberg said: "The number of compounds currently tested with this algorithm is small, and only further evaluation can let us know how good this method is and how accurate predictions are. This is indeed worth trying."