Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

Math Model Helps Predict Unknown Drug Side Effects

Aurel Cami (Harvard Medical School)

Aurel Cami (Harvard Medical School)

Researchers at Children’s Hospital Boston in Massachusetts have created a new method that combines data from a widely used drug safety database to predict adverse drug reactions. The findings from postdoctoral fellow Aurel Cami (pictured left) and colleagues from the Children’s Hospital Informatics Program, appear online in the the journal Science Translational Medicine (paid subscription required).

Existing side effect-detection methods often require a sufficient amount of adverse drug event (ADE) reporting to accumulate in clinical databases, a process that can take years. The databases are then subjected to data mining tools designed to detect previously unrecognized drug-ADEs relationships. Even with these processes, analysts may not be able to catch certain types of ADEs until patients have been on the drug for some time.

Cami’s team used snapshot from a drug safety database that captured descriptions of 852 side effects from 809 drugs, and created a formula known as a logistic regression predictive model. The model calculates the likelihood of unknown side effects of any drug in the network of almost one thousand compounds.

The team validated the model, called the Predictive Pharmacosafety Network, using current and historical drug safety data provided by Lexicomp, a database that collects information from drug package inserts. Based only on data available in 2005, for example, the model correctly identified 42 percent of the drug-ADE relationships that were subsequently discovered between 2006 and 2010.

One of those tests, again using only data available in 2005, was the relationship between the anti-diabetes drug rosiglitazone (marketed under the trade name Avandia) and heart attack. The model also correctly recognized as false fully 95 percent of drug-ADE pairs that in the 2010 data were categorized as having no association.

“We think the approach holds real promise for strengthening efforts to identify and manage drug risks,” says Cami, “by helping drug safety practitioners predict high likelihood events and guide efforts to understand, avoid, and alleviate those events before they start appearing in patients.” Cami adds they are now working on extending the model to add more drug safety data and promote its adoption in clinical drug safety practice.

Read more:

*     *     *

2 comments to Math Model Helps Predict Unknown Drug Side Effects