Recipient organisationRotherham Doncaster and South Humber NHS Foundation Trust
Funding£1.6M
PeriodOct 2023 — May 2027
In plain English
AI plain-English summary
A new AI tool will decide which psychological therapy a patient with depression receives first, rather than starting everyone on the least intensive option and waiting to see if it fails. This matters because NHS psychological services currently use a "stepped care" model: patients begin with low-intensity treatments, and only move to more intensive therapy if the first attempt doesn't work. That wastes time and resources for people who could have benefited from a different treatment from the start. The researchers have already shown in a trial of 951 patients that their AI-driven approach improves depression outcomes compared to usual care. Now they need to test whether it works across multiple services and is cost-effective over the long term. The trial will recruit 1,314 patients from 14 NHS psychological therapy services, randomly assigning whole services to either AI-guided treatment selection or standard stepped care. Patients will be followed for 18 months. If the AI approach proves effective and cost-effective, it could be rolled out across NHS mental health services, allowing clinicians to match patients to the right therapy on the first attempt rather than working through a fixed sequence.
View original technical description
Background: There are a number of empirically-supported psychological treatments for depression available in the NHS, but there is little evidence on how to decide which treatment may be most effective for each individual. Current practice in psychological therapy services is to offer treatments sequentially, usually starting with less intensive treatments followed by more intensive treatments for patients for whom the initial treatment fails (stepped care) [23]. Our research group has developed the first AI-driven decision support aid for stratified treatment selection using routinely-collected data from NHS psychological therapy services. We have tested its clinical effectiveness in a large (N=951) randomised controlled trial [13]. Our results indicate that stratified care is acceptable to implement in NHS services, significantly improving depression treatment outcomes by comparison to usual stepped care. However, questions remain about the generalisability and long-term cost-effectiveness of this approach. Aim: To test the clinical and cost effectiveness of AI-driven stratified care for depression symptoms in comparison to usual stepped care. Design: A pragmatic, multi-site, cluster randomised controlled trial (RCT). Target population & setting: We will recruit 1314 adult patients seeking treatment for common mental health problems and who present with case-level depression symptoms on the PHQ-9 measure [16]. They will be recruited from 14 psychological therapy services. Randomisation: Services (clusters) will be randomised 1:1 to either AI-driven stratified care or usual stepped care (control). Intervention: Participants will undergo an assessment after their referral for treatment, during which a treatment option will be selected. In the experimental group, treatment selection will be aided by an AI tool, while treatment selection in the control group will follow usual clinical practice and guidelines. Thereafter, participants in both groups will access their allocated psychological interventions and will be followed up at 6, 12 and 18 months. Outcomes: Depression symptoms measured by the PHQ-9 [16]. Secondary outcomes will include anxiety, quality of life. Data analysis: Quantitative, qualitative and health economic data analyses will be conducted. The primary analysis will compare depression outcomes between groups at 12 months, using mixed-effect regressions controlling for baseline severity. Patient & public involvement: Our study design is informed by consultation with our PPI Group. We will continue to involve our PPI Group in trial management, design of study materials, interpretation of qualitative data and dissemination of the findings. Team: Our multi-disciplinary team includes experts in clinical psychology, psychiatry, primary care, health services research, medical statistics, health economics, patient & public involvement, machine learning and AI. Impact: This trial will evaluate a new AI-driven approach to psychological treatment selection. This will allow us to plan and deliver more effective clinical services for this population, using a technology than can be easily scaled and efficiently implemented at a system-wide level in NHS psychological services.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know