Tech-mediated screening rivals psychiatrist interviews for youth mental health assessment

There is a shortage so severe that a single child psychiatrist must serve roughly 350 young people who need care. The wait for a psychiatric assessment can stretch months. A team of researchers in Slovenia decided to test whether a tablet chatbot or a humanoid robot could conduct the same screening interview a psychiatrist would—and whether the information gathered would be clinically useful.

The researchers recruited 106 young patients, ages 10 to 19, who were already receiving mental health treatment at a hospital in Maribor. They randomly assigned them to one of three groups: a face-to-face interview with a board-certified child psychiatrist, a self-administered screening on a tablet using a chatbot application, or an interview with a humanoid robot called Pepper. All three groups answered the same ten questions, designed to capture context, onset, and behavioral cues that complement standard screening tools. The questions were delivered conversationally, adapted to each modality.

When the young people rated their experience, they preferred the psychiatrist. Satisfaction with the interview itself was higher with the psychiatrist than with the chatbot. Satisfaction with communication—whether the interviewer seemed warm, listened well, understood them—was significantly higher with the psychiatrist than with either technology option. The robot and chatbot performed similarly to each other on these measures. But here is the critical finding: when asked whether they would be willing to do the screening again using the same method they had just experienced, the three groups showed no meaningful difference. Young people were equally willing to repeat a screening with a chatbot or robot as they were with a psychiatrist. This matters because it suggests that preference and practical acceptance are not the same thing. A teenager might prefer talking to a human, but when a human is not available, they will engage with the technology.

The researchers also looked at personality traits to see who was most satisfied with each modality. Two traits predicted satisfaction across all three conditions: conscientiousness—the tendency to be organized and reliable—and open-mindedness, or willingness to try new things. Young people who scored higher on these traits reported greater satisfaction with communication in the interview, regardless of whether they spoke to a psychiatrist, chatbot, or robot. This suggests that satisfaction is not purely about the technology itself but about how individual differences interact with it.

Next, the team showed video recordings of all 106 interviews to four child and adolescent psychiatrists and asked them to evaluate the clinical value of what they saw. The experts rated each interview on eight dimensions: whether it covered important psychosocial factors, whether it would help with diagnosis, whether it would support treatment planning, whether it could identify high-risk individuals, and whether it aligned well with other assessment methods. On nearly every dimension, the experts rated all three modalities favorably—above the midpoint of the scale. The only statistically significant difference was on alignment with other assessment methods, where the psychiatrist-led interviews scored higher than the robot interviews. The chatbot and psychiatrist interviews did not differ significantly on this measure. In other words, experts saw comparable clinical value in the information gathered by the chatbot and the psychiatrist, and only slightly less value in the robot's interviews when it came to fitting with existing assessment frameworks.

These findings suggest a path forward for youth mental health in places where psychiatrists are scarce. A young person could complete a screening interview on a tablet or with a robot, alone or with minimal supervision. The screening would gather clinically meaningful information—information that experts believe is as useful as what a psychiatrist would collect. That screening could then be reviewed by a clinician to decide whether the young person needs urgent care, ongoing treatment, or a follow-up appointment with a specialist. The technology would not replace psychiatrists; it would extend the reach of assessment into settings and times when psychiatrists are not available. It would also reduce the burden on the specialists who do exist, allowing them to focus on the young people who need their clinical judgment most.

The study has limits. It was conducted in a single hospital with young people already in treatment, which may mean they were more comfortable with all three modalities than youth who have never seen a mental health provider. The researchers did not measure whether young people actually disclosed more or less sensitive information to the technology versus the psychiatrist—only whether they were satisfied and willing to try again. And each expert evaluated videos independently; the researchers did not measure whether different experts would agree on the same interview. These gaps point to what comes next: larger studies across different settings, with youth who are not yet in care; measurement of actual disclosure patterns; and clearer protocols for how clinicians would use the screening information in real practice. The technology is also advancing rapidly. The chatbot used in this study was deliberately simple and rule-based, designed for safety and transparency rather than conversational flexibility. Future versions could incorporate more sophisticated language processing and adaptive questioning. But the core finding stands: a young person can complete a mental health screening with a machine and provide information that a psychiatrist will find clinically valuable.