Juan had been carrying the weight of depression for years, but when his psychiatrist asked how he was feeling, he always answered the same way: "Fine." That pattern, researchers say, is part of why mental health conditions like depression remain so difficult to diagnose and treat. But now, a team at the University of Southern California is developing a tool that could "listen" to the body instead of relying only on what patients say.
USC researchers have created an AI-powered system that can detect depression and suicidal thoughts by analyzing three types of biological signals: sweat, brain waves, and eye movements. The project, called PRECOG, started in June 2023 and has already produced five published studies this year, with another currently in the review process.
The team is led by Shrikanth Narayanan, who holds appointments at USC's Viterbi School of Engineering and the Stevens School of Computing and Artificial Intelligence. But Narayanan did not work alone. Engineers, neurologists, linguists, and psychiatrists all contributed their expertise to the project, making it a true collaboration across fields that rarely talk to each other.
One study looked at how people's eyes move when looking at different images. Another examined electrical activity in the brain using EEG, a method that places small sensors on the scalp. A third study tracked sweat gland activity, which changes when a person experiences emotional stress. In each case, the AI system learned to recognize patterns that distinguish people with depression or suicidal thoughts from those without.
The need for this kind of tool is urgent. According to the study, suicide rates in the United States have been climbing for decades. Among veterans, the rate is 1.5 times higher than in the general population. Currently, diagnoses of mental health conditions rely mainly on surveys and conversations between patients and doctors. But stigma and self-reporting biases often prevent people from sharing what they really feel.
Narayanan's team hopes their tool will give psychiatrists an objective way to identify people at risk, perhaps before they show obvious signs. The goal is not to replace human judgment but to offer doctors another signal to pay attention to, like a blood test for a physical illness. The technology could eventually help catch warning signs earlier, giving families and doctors more time to intervene.
The research is still in its early stages, and more testing with larger groups of people will be needed before the tool could be used in clinics. But for the millions of families touched by depression and loss, even a small step forward can feel meaningful. As the PRECOG project continues to grow, it offers a glimmer of hope that help may someday come before a crisis, not after.