Machine learning helps fine-tune drug prescribing for bipolar disorder
Mental health researchers at Harvard and the University of Pennsylvania have developed a novel machine learning technique for predicting how bipolar patients will respond to two drugs commonly prescribed to treat the disorder, according to a study running in Bipolar Disorders.
Calling their method elastic net regularization (ENR), the team described their work drawing data from 482 randomized outpatients who received talk therapy plus either lithium or quetiapine (brand name Seroquel).
They used ENR to build predictive models, then tested the predictions against a test set and measured patients’ clinical responses using an established scale for diagnosing bipolar disorder.
The predictive accuracy of the ENR model outperformed standard theoretical models, the authors reported.
Meanwhile the model’s predictions explained variances in clinical scores at the rate of 17.4% for the patients who took lithium and 32.1% for those treated with quetiapine.
The model accounted for such baseline variables as severity of mania, nonsuicidal self-injurious behavior, and preexisting phobias and anxiety conditions.
“Machine‐learning methods can identify predictors of response by examining variables simultaneously,” the authors wrote. “ENR is an effective approach for building optimal and generalizable models. Variables identified through this methodology can inform future research on predictors of response to lithium and quetiapine, as well as future modeling efforts of treatment choice in bipolar disorder.”
The study’s senior author was Andrew Nierenberg, MD, director of the Dauten Family Center for Bipolar Treatment Innovation at Massachusetts General Hospital.
Dave P. has worked in journalism, marketing and public relations for more than 30 years, frequently concentrating on hospitals, healthcare technology and Catholic communications. He has also specialized in fundraising communications, ghostwriting for CEOs of local, national and global charities, nonprofits and foundations.