Algorithm can predict which antidepressant will be more effective for patients
The path to an effective treatment for people with depression is slow, and often involves trial and error of several antidepressants before symptom relief happens, if at all. About 1 in 8 of the 242 million adults are currently prescribed an antidepressant medication, yet psychiatrists have no way to accurately predicting if a patient will benefit from a particular antidepressant medication. About two thirds of patients diagnosed with depression do not respond to the first antidepressant medication their doctor prescribes. Doctors typically prescribe antidepressants they are most comfortable with, yet a patient may need to wait 6 to 8 weeks before knowing if the medication is working for them, and have to repeat the process without knowing if they’ll feel better soon. When the risk of suicide and the ability to function is significantly impaired, a trial and error process is less than ideal.
Psychiatrists are currently flying blind to what is happening in the brain of the depressed patient, they can only assess behaviors but not the underlying cause of the disorder. A new study published in Nature Biotechnology reported that by using a depressed patient’s brain waves a team of researchers at Stanford University can now predict if a patient is likely to benefit from antidepressant medication.
The team at Stanford University led by Dr. Amit Etkin paired Hans Berger 1924 discovery of electroencephalogram (EEG) to measure brain activation in depressed patients and use Artificial intelligence (AI) and machine learning models to predict treatment response to the commonly prescribed antidepressant sertraline (Zoloft). EEG is a non-invasive technique that provides an insight into the distinct patterns of electrical activity across brain networks, and how these patterns of activation may be related to symptoms of depression.
- Introducing Open Mined: Decentralised AI
2. Only Numpy: Implementing Convolutional Neural Network using Numpy
3. TensorFlow Object Detection API tutorial
4. Artificial Intelligence Conference
Typically, about a third of depressed patient prescribed sertraline find relief from their symptoms of depression, when diagnosis is based on self-report. Etkin’s team reported that their AI algorithm is able to predict with 76% accuracy which patients are likely to respond to antidepressant medication when looking at their unique patterns of electrical activity in certain brain regions. Preliminary findings from other researchers is suggesting that patients with greater electrical activity on the left side of their brain are more likely to respond to sertraline.
The AI algorithm used by the researchers, was also able to predict which patients were more likely to benefit from non-pharmacological treatments for depression such as transcranial magnetic stimulation. The ability to tailor antidepressant treatment based on more than self-report measures, has long been a hope for psychiatry, and appears to be getting closer.
Brain science for teenagers. Neuroscience to empower teenagers on self-reliance, resilience, accountability and exuberance in children. www.teenbrain.info