The Magnificent Potential of Machine Learning

By: Melissa Liu

Machine learning is the usage of computer programs which automatically improve its algorithms. The program learns through experience by using statistics to find patterns in mass groups of data: a prominent subset of artificial intelligence (1).  For example, machine learning can detect patterns of specific diseases within “patient electronic healthcare records and inform clinicians of any anomalies” in a dataset of people with benign and malignant tumors of breast cancer (2). Machine learning in mental health has mainly been implemented to identify specific biomarkers, develop treatment plans, and predict crises (3). David Benrimoh, a psychiatry resident at McGill University, claims that machine learning algorithms will aid doctors in doing a “better job at determining relevant subtypes of different disorders and which treatments are most effective” to utilize. Through identifying sub-categories, doctors and mental health experts will “develop more catered treatment plans and medication dosages” (3), tailoring to precision medicine. 

For example, machine learning algorithms have been used for functional MRI (fMRI)brain images to identify patterns in  people with Major Depressive Disorder (MDD) compared with healthy controls (4). New York University’s Charles Marmar used neural networks to identify thirty voice features that may differentiate veterans with Post Traumatic Stress Disorder (PTSD) compared with unaffected veterans or healthy controls. Shorter vowel space has been shown by machine learning algorithms to be significantly altered in more than 250 individuals with PTSD (5). Not only does machine learning identify characteristics that humans are incapable for seeing , but machine learning algorithms have helped with an aspect of precision medicine in depression. Machine learning algorithms have been paired with clinical features to predict response to anti-depression medications, with an accuracy hovering at 60 percent – nothing too impressive yet, as this particular use of machine learning is still developing (6). 

Machine learning has also been used for the prevention aspect of precision medicine. Depression and suicide are strongly correlated: 60 percent of people who die by suicide have MDD (7). As suicide rates increase in the United States, especially over the past 30 years, there is a great need for preventative measures. Suicides accounted for more than 14 deaths every 100,000 people  in 2019 – or over 130 suicides per day (8,9). Studies show that there are thousands of risk factors that evoke suicide contemplation; one of the solutions to preventing these deaths is machine learning (10).  Studies from Vanderbilt and Florida Universities have found that a machine learning algorithm can accurately predict suicide attempts nearly 80 percent of the time, which is 20 percent higher than conventional logistic regressions of traditional risk factors (11). Even more so, the algorithm could be improved with more personal data about the individuals: life events, social media data, and career status – all aspects that are vital in developing the most precise and effective treatment for patients. Researchers at Carnegie Mellon found that machine learning algorithms could accurately detect neurosemantic signatures associated with suicide attempts – machine learning classified the brain images to find people with suicide risk (12). With the data collected from machine learning, precision medicine can be applied to deliver the right treatments to individuals who are at a high-risk of suicide or frequent depressive episodes.

The application of machine learning and electroencephalography (EEG) has also been effective and successful in designating medication in accordance with neuroscans. Due to the increase of evidence in pretreatment quantitative EEG measures, the information extracted will be useful for prediction of antidepressant response and remission for patients with MDD: Machine learning methods have been used to accurately predict EEG features predictive of symptom response to psychoactive drugs like clozapine (13,14,15,16). Machine learning has the potential to unravel pieces of  the complex battle with depression and allow scientists to learn more about  the most vulnerable people, thus prescribing them treatments which will be the most effective in response to their brain scans and neurological file. 

Listen to the audio: Audio

Contact me:

Artwork created by: Saavi Shinde


  1. Hao, Karen. “What Is Machine Learning?” MIT Technology Review, MIT Technology Review, 2 Apr. 2020, 
  2. Gharagyozyan, Hayk. “A Practical Application of Machine Learning in Medicine.” Macadamian, 10 Oct. 2019, 
  3. Abbas, Nabil M. “Machine Learning and Mental Health.” Medium, Towards Data Science, 5 Sept. 2019, 
  4. Franklin et al., “Risk Factors for Suicidal Thousands and Behaviors”. 2020
  5. Scherer, S., et al., “Self-Reported Symptoms of Depression and PTSD Are Associated with Reduced Vowel Space in Screening Interviews.” IEEE Transactions on Affective Computing, 2015. 7(1): pp 59-73
  6. Chekroud, A.M., et al., “Cross-Trial Prediction of Treatment Outcome in Depression: A Machine Learning Approach.” Lancet Psychiatry, 2016. 3(3): pp.243-250. 
  7. Ng CW, How CH, Ng YP. Depression in Primary Care: Assessing Suicide Risk. Singapore Med J. 2017;58(2):72-77. doi:10.11622/smedj.2017006
  8. Hutson, M., “Machine-Learning Algorithms Can Predict Suicide Risk More Readily Than Clinicians, Study Finds,” Newsweek. 2017
  9. “Suicide Statistics,” American Foundation for Suicide Prevention, July 19, 2019”. 
  10. Franklin, J.C., et al., “Risk Factors for Suicidal Thoughts and Behaviors: A Meta-Analysis of 50 Years of Research.” Psychol Bull, 2017. 143 (2): pp. 187-232; McConnon, A., “AI Helps Identify Those at Risk for Suicide,” Wall Street Journal. 2018. P. R7 
  11. Walsh, C.G. et al., “Predicting Risk of Suicide Attempts over Time Through Machine Learning.” Clinical Psychological Science, 2017. 5(3): pp.457-469. 
  12. Hutson, “Machine-Learning Algorithms Can Predict Suicide Risk.” Walsh et al., “Predicting Risk of Suicide Attempts.” 
  13. Pranav Rajpurkar, Jingbo Ying, Nathan Dass, Vinjai Vale, Arielle S. Keller, Jeremy Irvin, Zachary Taylor, Sanjay Basu, Andrew Ng, Leanne M. Williams, (2019). Evaluation of a Machine Learning Model Based on Pretreatment Symptoms and Electroencephalographic Features to Predict Outcomes of Antidepressant Treatment in Adults Within Depression. Jama Network Open, 2020. 
  14. Ferguson JM. SSRI antidepressant medications: adverse effects and tolerability. Prim Care Companion J Clin Psychiatry. 2001; 3(1): 22-27. 
  15. Khodayari-Rostamabad  A, Hasey  GM, Maccrimmon  DJ, Reilly  JP, de Bruin  H.  A pilot study to determine whether machine learning methodologies using pre-treatment electroencephalography can predict the symptomatic response to clozapine therapy.   Clin Neurophysiol. 2010;121(12):1998-2006. 
  16. Tenke  CE, Kayser  J, Manna  CG,  et al.  Current source density measures of electroencephalographic alpha predict antidepressant treatment response.   Biol Psychiatry. 2011;70(4):388-394. 

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s