A support vector machine neural pattern classifier differentiated between depressed patients who attempted suicide within the past 72 hours and those who were currently endorsing suicide ideation, according to a recent study published in the Journal of Clinical Psychiatry.

Using resting-state and task-based functional MRI (fMRI) to evaluate intrinsic brain activity associated with acute suicidal behavior, researchers were able to pinpoint patients who had recently attempted suicide (attempters) and those who had been harboring thoughts of suicide (ideators). The total inpatient and outpatient population consisted of 10 attempters, 9 ideators, 17 depressed nonsuicidal controls, and 18 nondepressed controls.

Based on data obtained from intake interviews and fMRI results, the binary classifier significantly discriminated attempters from ideators (mean accuracy: .788, P =.002). However, it was not able to discriminate recent attempters from all depressed patients (mean accuracy=0.531, P =.551) or clinically stable attempters from suicidal ideators (mean accuracy=0.58, P =.33).

Furthermore, researchers believe their classifier is specific to recent suicidal behavior, as it was not able to discriminate between presence vs absence of lifetime history of suicide attempts (mean accuracy=0.543, P =.348).

Nonimaging variables associated with acute suicide risk such as depression/anxiety severity and pressure pain threshold were found not to significantly discriminate attempters from ideators (mean accuracy=0.381, P =.236). They also did not improve the classifier’s accuracy (mean accuracy=0.736, P =.002).

Related Articles

There were several noted limitations to this study, such as lack of generalizability and inability to generate a sex-specific classifier, both of which were because of the small sample size.

Despite these and other limitations, researchers said, “the use of intrinsic brain activity to differentiate acute suicidal acts from current suicidal thoughts with greater accuracy than clinical measures is a promising step toward the development of an objective measure of imminent suicide risk.”


Cáceda R, Bush K, James GA, Stowe ZN, Kilts CD. Modes of resting functional brain organization differentiate suicidal thoughts and actions: a preliminary study [published online July 10, 2018] J Clin Psychiatry. doi:10.4088/JCP.17m11901