Summary: A new study demonstrates that an AI assistant can conduct psychiatric assessment interviews with greater diagnostic accuracy than widely used mental health rating scales. In a sample of 303 participants with confirmed psychiatric conditions, the AI assistant Alba provided DSM-based diagnostic suggestions after a brief conversational interview, outperforming rating scales in eight of nine disorders.
Alba was particularly effective at distinguishing conditions that often overlap, such as depression and anxiety, which traditional scales frequently score similarly. Participants also reported positive user experiences, indicating that conversational AI may offer a scalable, person-centered tool to support clinical assessment while preserving the clinician’s essential role.
A new study shows that an AI assistant can conduct assessment conversations with patients with higher accuracy than the rating scales used in healthcare today. In the study, 303 participants were interviewed by the AI assistant Alba, who then suggested possible psychiatric diagnoses.
In addition to being interviewed by an AI assistant, the participants also had to fill out standardized rating scales for the nine most common psychiatric diagnoses.
The results showed that the AI assistant’s assessments were more consistent with the participants’ actual diagnoses than the rating scales did.
The study included individuals with confirmed diagnoses for conditions such as depression, anxiety, obsessive-compulsive disorder, PTSD, ADHD, autism, eating disorders, substance use disorder and bipolar disorder, as well as a control group.
All participants had an online interview with the AI assistant, Alba, which asked 15-20 open questions about their mental health and then proposed diagnoses based on DSM-5 – the internationally-used manual for psychiatric diagnosis.
The AI assistant achieved higher accuracy in eight of the nine diagnoses, and could differentiate more clearly between diagnoses that often overlap. For example, the conventional rating scales often gave similar readings for depression and anxiety, whereas Alba’s assessments could discern the conditions more clearly.
The participants also described the user experience as positive – many perceived the AI assistant as empathic, relevant and supportive.
“An interview that can be done in a safe home environment before meeting a clinician has great value. The results point to a scalable, person-centred complement that can lighten the load for healthcare and provide a preliminary assessment, without replacing the psychologist or physician,” says Professor of Psychology Sverker Sikström, leader of the research team behind the study at Lund University and founder of the company, Talk To Alba.
Analyses the entire diagnostic manual – not just individual conditions
According to Sverker Sikström, the study marks a clear step forward in research on digital assessment tools for mental health. Previous studies have often been confined to analysing individual diagnoses or lacked clear justifications based on diagnostic criteria, whereas Alba can propose and justify all the diagnoses included in the DSM manual.
Fact box: What is Talk To Alba?
Talk To Alba is an online-based AI tool for the assessment, treatment and administration of mental health for professionals (psychologists, psychiatrists and physicians) who work in the area.
The tool includes AI-powered clinical interviews and CBT for patients, automatic diagnosis of mental health justified in accordance with DSM-5, informed AI dialogues about patients for clinics, and the transcribing and journal note-taking of patient meetings.
Alba, which is used at clinics in Sweden and abroad, is owned by TalkToAlba AB.
Generative AI-assisted clinical interviewing of mental health
The standard assessment of mental health typically involves clinical interviews conducted by highly trained clinicians. While effective, this approach faces substantial limitations, including high costs, high clinician workload, variability in expertise, and a lack of standardization.
Recent progress in large language models (LLMs) offer a promising avenue to address these limitations by simulating clinician-administered interviews through AI-powered systems.
However, few studies have rigorously validated such tools. In this study, we used TalkToAlba to develop and evaluat an AI assistant designed to conduct clinical interviews aligned with DSM-5 criteria.
Participants (N = 303) included individuals with self-reported clinician-diagnosed mental health disorders, namely, major depressive disorder (MDD), generalized anxiety disorder (GAD), obsessive-compulsive disorder (OCD), post-traumatic stress disorder (PTSD), attention-deficit/hyperactivity disorder (ADD/ADHD), autism spectrum disorder (ASD), eating disorders (ED), substance use disorder (SUD), and bipolar disorder (BD) — alongside healthy controls.
The AI assistant conducted diagnostic interviews and assessed the likelihood of each disorder, while another AI system analyzed interview transcripts to verify diagnostic criteria and generate comprehensive justifications for its conclusions.
The results showed that the AI-powered clinical interview achieved higher agreement (i.e., Cohen’s Kappa), sensitivity, and specificity in identifying self-reported, clinician-diagnosed disorders compared to established rating scales.
It also exhibited significantly lower co-dependencies between diagnostic categories. Additionally, most participants rated the AI-powered interview as highly empathic, relevant, understanding, and supportive.
These findings suggest that AI-powered clinical interviews can serve as accurate, standardized, and person-centered tools for assessing common mental disorders.
Their scalability, low cost, and positive user experience position them as a valuable complement to traditional diagnostic methods, with potential for widespread application in mental health care delivery.
Read more on Neuroscience News
