After honoring our veterans this past Veteran’s Day, it is also important to shed light on various ways to help ensure that Veterans and others with mental illnesses can receive the care that they may need for scars both visible and invisible.  Sadly, one of the most affected groups of suicide are veterans with an average of 20 Veterans passing away each day due to suicide, according to the Department of Veterans Affairs.

As written in a previous SIIA AI Spotlight, in the United States, suicide is ranked as the third leading cause of death among people ages 10 to 14, second among people ages 15 to 24, fourth among people ages 35-54, and tenth overall according to the Center for Disease Control and Prevention.  One such tool that may aid in the field of suicide prevention is, unsurprisingly, artificial intelligence (AI) and machine learning.

Researchers from Carnegie Mellon University and Harvard University developed a machine learning algorithm that, when paired with an MRI machine, can detect for suicidal behavior using brain scans.  The findings of this study showed that there is a detectable difference in the brains of people who contemplate suicide than those who do not, something that was not previously proven.  This breakthrough will hopefully enable doctors and researchers to progress and find new ways to detect and treat patients and other persons who are at risk of attempting suicide.

The researchers used 34 patients for their MRI test.  17 had reported suicidal thoughts and the other 17 did not.  They presented the patients with the following keywords: death, cruelty, trouble, carefree, good, and praise.  The algorithm that the researchers developed correctly identified 15 out of 17 of the patients with suicidal thoughts and 16 out of 17 of the control group. 

This 91 percent accuracy rate shows great promise for mental health treatment, though it still has some limitations.  For one, patients would have to agree to receive an MRI, and a patient who is contemplating suicide might not want to subject himself or herself to an invasive and expensive process.  However, these findings show a core difference in the brains of people who are contemplating suicide versus those who are not which could improve detection and prevention early on.

Harvard and Carnegie Mellon are not the first universities or teams of researchers to work to use AI for suicide prevention.  In fact, there are ongoing efforts to use AI to improve treatment and response for people who suffer from mental illnesses.  As written previously, Facebook has also been implementing AI tools to help detect users who may be contemplating suicide and can also provide these users with tools to use to seek help.  Additionally, Vanderbilt University Medical Center data scientists developed algorithms to predict if someone is at risk of attempting suicide with an accuracy rate of 80-90 percent compared to the just over 50 percent rate of doctors who don’t use AI tools.  The Harvard University and Carnegie Mellon University research collaboration is another extension of this exceptionally important work.

A doctor who is scanning a patient with known or unknown mental illness using an MRI may now have another tool to refer to when analyzing a brain scan.  AI in this capacity is just that, a tool, not a replacement for doctors who are still exceptionally important to the interpretation and diagnostic process.  This is in line with what SIIA has previously written in an issue brief on AI and the future of work.  AI is largely used as a tool to supplement and enhance human work rather than replace it.

AI as a tool for suicide prevention certainly shows promise, though is by no means a cure-all. However, as research has shown that doctors alone only have an accuracy rating of just over 50 percent when predicting if someone is about to attempt suicide, AI provides a substantial boost making the diagnoses far more accurate.  Hopefully, there will be promising results and people suffering from depression and other mental illnesses will be able to seek and receive much needed help.

Diane Pinto is the Public Policy Coordinator at SIIA. Follow the Policy team on Twitter @SIIAPolicy.