New research suggests artificial intelligence might one day help clinicians spot depression, anxiety and stress earlier — simply by analyzing standard questionnaire data.
In a new study out of Nature Scientific Reports, researchers looked at answers from nearly 40,000 people who took a well-known mental health questionnaire called the DASS-42. It’s something psychologists use to measure depression, anxiety and stress. The researchers then trained different AI models to look for patterns in those answers — basically teaching the computer what “risk” looks like in real people’s responses.
Researchers said it worked surprisingly well. One model in particular — called a support vector machine — was right more than 98% of the time across all three conditions. The support vector machine model was right 99.3% for depression, 98.9% for anxiety and 98.8% for stress.
“Artificial intelligence (AI) has proven its use in healthcare applications in which performing tasks becomes easier and more efficient for detection and diagnosing of diseases, drugs development and predictive analysis. The improvement in collecting and analyzing data from images, voice records, videos, posts, and interaction over social media platforms made AI very useful in mental healthcare,” the study stated.
The figure is incredibly high for mental health research, and it suggests AI can pick up subtle signals that might be missed by humans, especially in large groups of people.
“Open-text fields required additional preprocessing, which may introduce standardization errors. Future research may explore deep learning approaches and integrate multimodal data sources like voice or physiological signals. Validating these models in clinical settings would also strengthen their applicability in real-world healthcare environments. The predictive ability of AI-based models such as SVM highlights their potential role in real-time clinical settings,” the study stated.
Researchers say this technology could have significant impact on everyday mental health care.
“AI technology has impacted mental healthcare in various ways such as collecting data about the patient through photos, videos, music they listen to, posts and interactions through social media, and information from wearable devices such as smart watches, after collecting these data, machine learning (ML) algorithms predict the individual’s mental health by analyzing data collected,” the study stated.
Right now, diagnosing depression or anxiety usually requires time-intensive interviews with trained professionals. If AI can reliably flag people at risk from simple questionnaires, it could be used as a screening tool — helping to prioritize who needs care faster. The main takeaway here: artificial intelligence may soon become another tool in the mental health toolbox — not to diagnose, but helping to flag concerns earlier.
Researchers say the next step is testing these systems in real clinical settings.
“Validating these models in clinical settings would also strengthen their applicability in real-world healthcare environments. The predictive ability of AI-based models such as SVM (support vector machine) highlights their potential role in real-time clinical settings,” the study stated.