In recent years, the U.S. has observed a troubling increase in suicide rates, with veterans experiencing rates that are 1.5 times higher than the general population, according to research from the USC Suzanne Dworak-Peck School of Social Work. Effective intervention plays a vital role in suicide prevention, but many incidents, especially within military circles, remain unreported due to societal stigma, mental health misconceptions, and biases in disclosure.
A significant challenge in addressing these issues is the absence of standardized testing methods for mental illnesses like depression. Currently, diagnoses primarily depend on self-reported assessments, including surveys and clinical consultations, making objective evaluation crucial.
To tackle this pressing need, USC researchers—specializing in fields ranging from engineering to neurology and artificial intelligence—have initiated a groundbreaking study. They aim to assess depression and suicidal ideation through the analysis of various biomarkers, including sweat, brain signals, and eye movements, using AI for classification. Led by Shrikanth Narayanan, who holds several distinguished appointments at USC, this initiative seeks to provide clinicians with reliable tools for identifying individuals at high risk based on biological indicators, enhancing the precision of mental health monitoring and intervention.
The study, named "PRECOG: Multimodal integration of neural and biobehavioral signals for predicting preconscious responses," began in June 2023 and has already produced five published papers in reputable journals, with another in progress that discusses additional biomarkers and computational advancements.
One highlighted paper published in npj Digital Medicine, titled "Deep Learning Characterizes Depression and Suicidal Ideation in Young Adults From Eye Movements," was spearheaded by Ph.D. student Kleanthis Avramidis. Additionally, research focusing on neural signals derived from electroencephalography (EEG) has resulted in three significant publications covering unique insights into depression and suicidality from various Ph.D. students, each exploring different neural dynamics through advanced EEG tracking.
The team has also utilized electrodermal activity (EDA) to gather data on skin conductance, linking it to emotional responses, thereby reinforcing the connection between physiological reactions and mental health conditions. Avramidis led one investigation into this area, the findings of which are available on the arXiv preprint server.
The PRECOG project thus aims to shift mental health diagnoses towards a more evidence-based methodology, grounded in physiological data rather than solely depending on subjective assessments. Among the biomarkers studied, researchers have utilized EEG signals, EDA, and eye-tracking to construct AI and deep learning models aimed at identifying stable indicators of depression. By observing how biological systems react to emotional prompts, the research focuses on quantifying the biological aspects of thoughts and feelings.
The role of language also features prominently in this study, as researchers deployed a sentence evaluation task where individuals processed 160 self-referential statements with varying emotional tones. Participants’ brain activity, eye movements, and skin conductance responses were monitored, offering a comprehensive look at their emotional responses to these statements.
EEG technology was employed to record real-time neural responses as participants engaged with the emotionally charged language. The analysis revealed distinct patterns that differentiate healthy individuals from those with depression or suicidal ideation, honing in on specific time frames where brain responses diverged significantly.
Moreover, the team studied eye movements—considered an involuntary action—as a potential measure of cognitive processing. Through an advanced eye-tracking system, they analyzed how gaze behavior varied between healthy participants and those expressing suicidal thoughts while interacting with the same self-referential sentences.
Finally, researchers examined sweat as a marker of emotional response through EDA, with findings suggesting that physiological changes linked to negative stimuli could act as valuable biomarkers for depression and suicidal tendencies.
By incorporating these innovative methods, the researchers aspire to enhance clinical mental health assessments. As emphasized by Byrd, the goal is not to replace existing assessments but to complement them, allowing mental health professionals to make more data-informed decisions and precisely identify intervention opportunities for those at risk of suicidal ideation.
