AI technology predicts whether patients will benefit from an antidepressant

Depression (most commonly major depressive disorder, or MDD) is one of the most common mental illnesses in the world, with an estimated 264 million people around the globe suffering from at least one form of this category of mental illness. Anywhere from 10-30% of these people suffer from treatment-resistant depression, meaning that they do not have a significant positive response to one or more antidepressant treatments or medications. However, a new technology developed by researchers at California’s Stanford University may pave the way for a new diagnostic strategy that will help patients get the type of treatment they need without going through months of hurdles and trials.

MDD can severely impair sufferers’ ability to carry out major life functions, and patients often struggle to cope with stress from relationships and careers. It is often misdiagnosed as other mental conditions, such as bipolar disorder, which can interfere with patients getting effective treatment. Not only is ineffective medication a waste of money and time, but it also may cause adverse side effects in patients, including an increase in depressive or suicidal thoughts.

“We have a central problem in psychiatry because we characterize diseases by their end point, such as what behaviors they cause. You tell me you’re depressed, and I don’t know any more than that. I don’t really know what’s going on in the brain and we prescribe medication on very little information.”

Amit Etkin, PhD, Study Co-Author and Professor of Psychiatry and Behavioral Sciences at Stanford University

Amit Etkin, lead author of the Stanford study, wanted to see if an AI could predict an antidepressant’s effectiveness by reading patient brain waves before they had even received treatment. The study focused on the drug sertraline, which is effective in about 1/3 of patients with depression.

Sertraline is a type of drug known as a selective serotonin re-uptake inhibitor, or SSRI. It works by affecting levels of serotonin in the brain, as the name suggests. Serotonin is a neurotransmitter (chemical messenger) that travels between nerve cells in the brain. Sometimes referred to as the “happy chemical”, it influences mood, emotion, and sleep cycles. After a message is received, typically the serotonin molecule is reabsorbed and recycled by nerve cells in a process known as re-uptake. However, SSRI medications block this re-uptake, meaning the serotonin’s message continues on and affects more nerve cells, effectively increasing serotonin levels across the brain.

While it would be too simplistic of an explanation to say that depression (and other related mental health conditions) are caused by low serotonin levels, but several studies have shown that a rise in serotonin levels can improve symptoms. Higher levels of serotonin can also make sufferers more responsive to other types of treatment, including therapy and different classes of antidepressants. As a result, sertraline is used to treat a multitude of mental disorders other than depression, including obsessive-compulsive disorder (OCD), post-traumatic stress disorder (PTSD), and panic disorder.

Etkin’s team used an electroencephalogram (EEG) machine to analyze the brainwaves of 228 people suffering from depression aged 18-65 who were not actively using antidepressants. Half of the participants were then given sertraline and the other half took a placebo. The researchers then monitored each participant’s moods over the next 8 weeks. The machine learning-algorithm then compared EEG readings of patients who responded either well or poorly to sertraline. The AI found a specific pattern of brain activity that was correlated with a higher likelihood of finding sertraline helpful. When the AI was tested on a different group, the AI had a 76% success rate of predicting which people would benefit from the drug.

Etkin has founded a company called Alto Neuroscience to further develop this AI. However, more trials will be needed before this technology becomes a mainstream diagnostic tool.

If you or a loved one are experiencing suicidal thoughts, contact the National Suicide Crisis Line at 1-800-273-8255.

Original story and interview by Jason Arunn Murugesu.

References

1. Wu W, Zhang Y, Jiang J, Lucas MV, Fonzo GA, Rolle CE, Cooper C, Chin-Fatt C, Krepel N, Cornelssen CA, Wright R, Toll RT, Trivedi HM, Monuszko K, Caudle TL, Sarhadi K, Jha MK, Trombello JM, Deckersbach T, Adams P, McGrath PJ, Weissman MM, Fava M, Pizzagalli DA, Arns M, Trivedi MH, Etkin A. 2020. An electroencephalographic signature predicts antidepressant response in major depression. Nature Biotechnology.
2. Murugesu JA. 2020. Brain activity can help predict who’ll benefit from an antidepressant. NewScientist. New Scientist Ltd.
3. GBD 2017 Disease and Injury Incidence and Prevalence Collaborators. 2019. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. The Lancet 392:e44.
4. Al-Harbi KS. 2012. Treatment-resistant depression: therapeutic trends, challenges, and future directions. Patient Prefer Adherence 6:369–388.
5. National Health Services. 2019. SSRIs (selective serotonin reuptake inhibitors). NHS Direct Wales. Welsh Ambulance Services NHS Trust.
6. Acharya UR, Sudarshan VK, Adeli H, Santhosh J, Koh JEW, Adeli A. 2015. Computer-Aided Diagnosis of Depression Using EEG Signals. European Neurology 73:329–336.