New project explores AI as a tool for identifying mental health conditions

Can AI accurately detect the symptoms of mental health conditions? And could this be used to improve support for people who are affected by them? This is the focus of a new collaborative research project between Spicy Minds and the University of Bristol.

Students from the university’s School of Physiology, Pharmacology and Neuroscience are testing the capabilities of AI using Rowan – the app developed by Spicy Minds that gives mental health support that incorporates evidence-based therapeutic techniques delivered by AI.  

The aim is to see if the AI can be trained to identify the range of symptoms that typically present in a set of mental health conditions, which have been carefully selected because they have overlapping symptoms. Boundaries between distinct conditions are not clear cut but some have symptoms that connect them. This project will explore the role of AI in improving how support is given across different conditions – known as a transdiagnostic approach. 

The project will require the final-year neuroscience students to create different AI ‘characters’ which will be archetypes of a particular mental health condition: generalised anxiety disorder, post-traumatic stress disorder (PTSD), addiction, attention deficit hyperactivity disorder (ADHD) and a neurotypical character who is facing day-to-day stresses. 

Rowan will then have conversations with these ‘characters’, offering them support with the particular imagined challenges that they are struggling with.

The students will then analyse the transcripts from the conversations to discover if Rowan was able to identify symptoms of mental health conditions and use appropriate and effective techniques to help the ‘character’. They will also examine whether Rowan can detect overlapping symptoms across conditions in the conversations.

Dr Emma Cahill, lecturer in neuroscience at the University of Bristol, said:

“My work to date has focused on the behaviour of animals and how we can use that to understand the human brain. Currently, mental health conditions are identified using diagnostic criteria informed by animal modelling of specific symptoms. I believe that a focus on identifying shared symptoms across conditions will stimulate new ideas for treatments.

“That’s why this project is so interesting and could have wide ranging implications for the world of psychology and psychiatry. If we can show that AI can be trained to reliably identify symptoms and offer support, it could be a powerful new tool for mental health professionals as well as clients.”

The project is part of the university’s programme of Community Engagement Learning, which gives students across a range of subjects the opportunity to apply skills and knowledge from their studies to a real-world issue.

The students will also test Rowan by introducing a moment of crisis into the conversations to see how it handles the situation. This could be the addiction ‘character’ telling Rowan that they have taken drugs or the PTSD ‘character’ saying they had a flashback.

Ben Cosh, CEO of Spicy Minds, says: “I am hugely excited by what we are creating with Rowan. This is advanced AI that is able to provide incredibly nuanced and compassionate mental health support. This partnership with the University of Bristol is not only giving students valuable exposure to real world solutions but also means Rowan’s capabilities are being tested with academic rigour. We can’t wait to see the outcomes in the spring.”  

The project runs until March, and results of the research will be available in spring 2025.