Researchers at Columbia University have developed a machine learning algorithm that identifies and predicts gender-based differences in adverse reactions to drugs, according to a report published online Sept. 22 in Patterns.
Dubbed AwareDX–Analyzing Women At Risk for Experiencing Drug toxicity, the algorithm automatically corrects for the biases in drug effects data that stem from overrepresentation of male subjects in clinical research trials.
Nicholas Tatonetti, PhD, and colleagues created the algorithm based on the observation that, while men and women can experience different side-effects from certain medications, physicians may not be aware of them because most clinical trial data are biased toward men. This, they note in the report, impacts prescribing guidelines, drug marketing and, ultimately, patients’ health.
The algorithm leverages 52 years of data from the FDA’s Adverse Event Reporting System (FAERS), which contains reports of adverse drug effects from consumers, healthcare providers and manufacturers. Tatonetti and co-author Paydal Chandak, a senior biomedical informatics major at Columbia, compiled the data into a bank of more than 20,000 potential gender-specific drug effects.
These effects can be verified by looking back at older data or conducting new studies. The algorithm groups data into gender-balanced subsets before looking for patterns and trends, repeating the search process 25 times to improve search results.
The researchers note that, while there is significant work left to do, they have already had success verifying the results for several drugs based on previous research.
For example, they hypothesized that men who take the cholesterol medication simvastatin are more likely than women to be at risk for muscle aches. They also theorized that women who are prescribed the antipsychotic medication risperidone are at greater risk of a slower heart rate than men who take the same drug.
Both hypotheses were based on the fact that the ABCB1 gene, which impacts how much of a drug is usable by the body and for how long, is known to be more active in men than in women. The algorithm successfully predicted the effects of simvastatin and risperidone, the authors write.
“The most exciting thing to me is that not only do we have a database of adverse events that we’ve developed from this FDA resource, but we’ve shown that for some of these events, there is preexisting knowledge of genetic differences between men and women,” Chandak adds in prepared remarks.
The researchers hope that continued efforts to verify their results will mean physicians use insights gleaned from the algorithm to make more informed choices when prescribing drugs, especially to female patients.