A new study has found that an AI assistant can conduct mental-health interviews with greater accuracy than the rating scales commonly used in healthcare. In the study, 303 participants spoke with an AI assistant named Alba. After each interview, Alba suggested possible psychiatric diagnoses.
All participants also completed standardized rating scales that measure the nine most common psychiatric conditions. When researchers compared the results, they discovered that Alba’s assessments aligned more closely with the participants’ confirmed diagnoses than the traditional scales did.
The study included people already diagnosed with conditions such as depression, anxiety, obsessive-compulsive disorder, PTSD, ADHD, autism, eating disorders, substance use disorder and bipolar disorder. A control group was also included.
Each participant completed an online session with Alba, which asked 15–20 open-ended questions about their mental health. Alba then proposed diagnoses using the DSM-5, the global standard for psychiatric classification.
The AI assistant delivered higher accuracy in eight of the nine diagnostic categories. It also distinguished more clearly between conditions that often overlap. For instance, conventional scales often produce similar results for depression and anxiety, but Alba separated the two more effectively. Participants described the experience as positive and said Alba felt empathetic, relevant and supportive.
“An interview done safely at home before meeting a clinician has great value,” says Sverker Sikström, Professor of Psychology at Lund University and leader of the research team. He explains that the AI tool can serve as a scalable and person-centred support that eases workloads for clinicians while offering an early assessment. However, he stresses that it is not meant to replace psychologists or physicians.
Analyses Based on the Full Diagnostic Manual
Sikström says the findings represent a major step forward for digital mental-health assessment. Earlier tools often focused on single conditions or lacked clear explanations tied to diagnostic criteria. In contrast, Alba can propose and justify every diagnosis listed in the DSM.
What Is Talk To Alba?
Talk To Alba is an online AI system used by mental-health professionals, including psychologists, psychiatrists and physicians. It provides AI-driven clinical interviews, CBT-based support for patients, automated DSM-5-aligned diagnoses, AI-assisted clinical dialogues and transcription of patient sessions. The system is already in use at clinics in Sweden and other countries and is owned by TalkToAlba AB.
Related topics
