A new AI‑driven tool that tailors antidepressant treatment to individual patients could significantly improve outcomes for people with depression, according to major new research led by the University of Oxford.
The tool, known as PETRUSHKA, is the first mental health clinical prediction tool shown to be effective in real‑world trials. Designed to support shared decision‑making between clinicians and patients, it uses artificial intelligence to recommend the antidepressant most likely to work for each individual while accounting for personal preferences and potential side effects.
PETRUSHKA was developed through an NIHR Research Professorship, with additional support from the NIHR Oxford Health Biomedical Research Centre and the NIHR Applied Research Collaboration Oxford and Thames Valley.
The tool combines clinical history, demographic information, and patient preferences, including side‑effect concerns.
It takes around three minutes to use during a consultation and was tested in a large randomised clinical trial involving 500+ adults with major depressive disorder across the UK, Brazil and Canada. PETRUSHKA launched in 2024 as part of the trial.
Published in the Journal of the American Medical Association, the findings show that patients whose antidepressant was selected using PETRUSHKA were:
- Around 40% less likely to discontinue treatment within the first eight weeks
- More likely to experience better mental health outcomes over six months
- Less likely to stop medication due to side effects
- Showing greater improvements in depression and anxiety by 24 weeks
Researchers say the tool could reduce a major challenge in depression treatment: early discontinuation of medication due to side effects or lack of perceived benefit.
Crucially, PETRUSHKA was developed with people who have lived experience of depression, ensuring that the tool reflects the priorities, concerns and real‑world decisions faced by patients.
It is specifically designed for use in everyday clinical settings, including primary care, where most antidepressants are prescribed. Clinicians and patients use the tool together in a consultation, supporting personalised, evidence‑based and preference‑informed decisions.
Depression remains one of the most common and costly health conditions worldwide, contributing to:
- Lost productivity
- Pressure on primary care and mental health services
- High rates of treatment switching and discontinuation
Researchers say PETRUSHKA could help address these issues by improving treatment adherence and symptom improvement.
However, they stress that further research is needed into long‑term outcomes, impact on NHS resources and broader cost‑effectiveness before the tool is widely adopted across the health system.
Mike Lewis, NIHR Scientific Director for Innovation, commented:
“This study shows the real power of combining digital technology and personalised treatment.
"By harnessing data and embracing cutting-edge digital tools, we can tailor care more precisely to each patient - improving outcomes for individuals. Forward-thinking research such as this shows the impact of using innovative technologies to ensure patients across the UK receive smarter, more effective care.”
For NHS and public sector leaders, PETRUSHKA represents a significant milestone in the growing integration of AI in mental health. It demonstrates how data‑driven tools can support clinicians, improve patient experience and enhance treatment outcomes without replacing the human relationship at the heart of care.
The research team hopes the findings will pave the way for wider adoption of intelligent decision‑support tools in mental health settings — an area where innovation has historically lagged behind other clinical specialties.
Image credit: iStock
