AI-Powered Systems Deployed to Detect and Prevent Suicide Risk in Multiple Settings

Aug 2, 2026

To address rising suicide rates in the United States, University of Southern California researchers launched the PRECOG project in June 2023. The study utilizes artificial intelligence to analyze biological signals—including brain activity via 64-channel EEG, eye movements, and skin conductance—to identify objective markers of depression and suicidal ideation. Rather than replacing traditional self-reports, these metrics aim to provide clinicians with data-driven intervention points.

PRECOG multimodal biomarker study

▪The PRECOG project resulted in five published papers in major journals in 2026, with one additional paper in the publication process.
▪The PRECOG study analyzed biomarkers including neurological signals, electrodermal activity, and eye-tracking behavioral indicators to identify stable markers of depression.
▪University of Southern California researchers launched the PRECOG project in June 2023 to classify depression and suicidal ideation using biological signals and artificial intelligence.

Rising U.S. suicide rates

▪Traditional mental health diagnoses in the United States rely primarily on self-reporting tools such as surveys and clinical interviews due to a lack of standard objective testing methods.
▪Suicide rates in the United States have risen over the past few decades, with rates among veterans being 1.5 times higher than the general public.

EEG neural signature findings

▪Researchers used a 64-channel electroencephalography system to record electrical brain activity while participants read 160 self-referential statements.
▪Researchers identified a blunting of the N170 brain response that tracks the severity of suicidal ideation and distinguishes suicide risk from general depression.
▪Time-resolved analyses of electroencephalography data showed the most reliable differences between healthy, depressed, and suicidal groups emerged 300 to 600 milliseconds after word presentation.

Eye-tracking depression patterns

▪Researchers used an infrared EyeLink 1000 Plus eye-tracking system to record participants' eye movements during a sentence evaluation task.
▪Deep learning analysis revealed that horizontal gaze patterns carried the most diagnostic information for identifying depression and suicidal ideation, particularly during negative statements.
▪Individuals with suicidal ideation exhibited dispersed or disengaged viewing patterns when processing negative statements, whereas healthy participants showed structured visual patterns aligned with the disagree option.

Electrodermal activity measurement

▪Researchers recorded skin conductance responses to measure physiological arousal through the sympathetic nervous system as participants processed emotionally charged words.
▪Computational modeling of skin conductance showed that physiological reactions to negative words carried the strongest diagnostic information for depression and suicidal ideation.

Clinical assessment supplementation goals

▪The PRECOG study aims to supplement, rather than replace, existing clinical mental health assessments to help psychiatrists make data-driven decisions.
▪Objective biological testing metrics are designed to provide clinicians with physiological data to identify precise intervention points for patients experiencing suicidal ideation.