Following on from prior research suggesting particular brain wave patterns can signal whether or not antidepressant medication will be effective, a new study is reporting the development of an algorithm that can automatically predict patient response to antidepressants based on simple EEG data.
Back in 2011 a large, multi-center project was launched called EMBARC (Establishing Moderators and Biosignatures of Antidepressant Response for Clinical Care). The goal of the project was to develop a set of objective diagnostic biomarkers to help clinicians better evaluate, and treat, patients with depression.
Madhukar Trivedi, a professor of psychiatry from UT Southwestern, explains the goal of EMBARC was to overcome the current trial-and-error process clinicians currently move through when trying to find an appropriate treatment for patients suffering from depression.
“It often takes many steps for a patient with depression to get better,” says Trivedi. “We went into this thinking, ‘Wouldn’t it be better to identify at the beginning of treatment which treatments would be the best for which patients?’”
The EMBARC project recruited around 300 subjects with depression, and randomly assigned them to either a placebo or the antidepressant sertraline for eight weeks. One of the data points gathered by the study involved taking EEG measurements at the beginning of the trial to investigate whether a simple brain wave test could predict which patients were most likely to benefit from antidepressant medications.
Early evaluation of the EMBARC data suggested antidepressant success could be predicted using EEG data. This new study turns that evidence into a practical clinical test demonstrating the development of an effective predictive model based on the initial data.
The machine learning model developed is called SELSER (Sparse EEG Latent SpacE Regression). The SELSER model was subsequently tested in several independent data sets and found to outperform all other predictive models currently available to clinicians.
As well as detecting patients most likely be benefit from sertraline treatments, the SELSER model suggests patients unlikely to improve from antidepressants were more likely to respond to alternative treatments such as transcranial magnetic stimulation.
“This study takes previous research showing that we can predict who benefits from an antidepressant and actually brings it to the point of practical utility,” says Amit Ekin, from Stanford University and co-senior author on the new research. “Using this method we can characterize something about an individual person’s brain. It’s a method that can work across different types of EEG equipment, and thus more apt to reach the clinic.”
Other clinical tests for depression are still in development based on the EMBARC trial data, including blood biomarkers and MRI scans. However, the cheap and prolific nature of EEG scans make this a most promising tool for easy and rapid deployment into clinical environments. More work is certainly needed before the new test is widely available, but Ekin is confident the technology is on track to move into clinical uses reasonably soon.
“I will be surprised if this isn’t widely used by clinicians within the next five years,” he suggests.
The new study was published in the journal Nature Biotechnology.
Source: Stanford Medicine/UT Southwestern/NIH
