Breakthrough in Precision Psychiatry: Oxford's PETRUSHKA Tool Leads the Way

Major depressive disorder affects millions in the United Kingdom, with around 89 million antidepressant prescriptions issued in 2023/24 alone, marking a steady rise. Yet, traditional trial-and-error prescribing often leads to high dropout rates—up to 30% within the first weeks—due to side effects or lack of efficacy. A groundbreaking study from the University of Oxford, funded by the National Institute for Health and Care Research (NIHR), introduces PETRUSHKA, an AI-driven decision-support tool revolutionising antidepressant personalisation. This innovation promises better outcomes for patients while easing the burden on NHS clinicians.

Developed at Oxford's Precision Psychiatry Lab, PETRUSHKA combines vast clinical trial data, real-world evidence, and individual patient factors to recommend tailored treatments. The results, published in the Journal of the American Medical Association (JAMA), mark the first robust evidence that such a tool enhances mental health care in routine settings.

The Scale of Depression in the UK: Why Personalisation Matters

Depression is the leading cause of disability worldwide, impacting over 6 million adults in England yearly. In the UK, one in six people experiences it at some point, with women twice as likely as men. Antidepressants like SSRIs (Selective Serotonin Reuptake Inhibitors) are first-line treatments, but only about 30-40% respond fully on the initial choice, leading to prolonged suffering and healthcare costs exceeding £12 billion annually.

Current prescribing relies on clinician judgment and guidelines, but variability in patient response—driven by genetics, demographics, and preferences—means many discontinue early. PETRUSHKA addresses this by shifting from one-size-fits-all to precision approaches, aligning with NIHR's push for innovative mental health tools.

What is PETRUSHKA? A Deep Dive into the AI Tool

PETRUSHKA stands for Personalising antidEpressant Treatment foR Unipolar depreSsion combining individual cHoices, risKs and big datA. Co-produced with patients who have lived experience of depression, this web-based tool takes just three minutes to generate recommendations. It integrates data from over one million patients, weighing factors like age, gender, symptom severity (via PHQ-9 scores), comorbidities, and crucially, patient priorities such as avoiding specific side effects like weight gain or sexual dysfunction.

How PETRUSHKA Works: Step-by-Step Process

The tool's AI employs statistical and machine learning models for predictions. Here's the process:

• Input Collection: Clinician enters patient details (demographics, symptoms) and patient shares preferences via questionnaire.
• Data Integration: Draws from randomised trials, observational studies, and network meta-analyses for efficacy/tolerability rankings.
• Prediction Modelling: Ranks antidepressants by expected response and side effect risks, prioritising patient choices.
• Output Delivery: Provides top recommendations with probabilities, supporting shared decision-making.
• Integration: Usable in primary care or remotely, fitting busy NHS workflows.

This patient-centred design distinguishes PETRUSHKA from biomarker-heavy tools, making it practical for UK universities and clinics researching scalable interventions.

The PETRUSHKA Trial: Rigorous Testing Across Continents

Launched in late 2022, the PETRUSHKA randomised controlled trial (RCT) spanned 47 sites in the UK, Brazil, and Canada, enrolling 540 adults aged 18-74 with moderate-to-severe depression. Participants were randomised 1:1 to PETRUSHKA-guided prescribing or usual care.

Primary endpoint: discontinuation at 8 weeks. Follow-up extended to 24 weeks, measuring PHQ-9 (depression) and GAD-7 (anxiety) scores. Led by Professor Andrea Cipriani, NIHR Research Professor at Oxford's Department of Psychiatry, the trial exemplifies collaborative academic research.Read the full Oxford announcement.

Photo by Markus Winkler on Unsplash

Game-Changing Results: Fewer Dropouts, Better Symptom Relief

At 8 weeks, only 17% in the PETRUSHKA arm discontinued (vs 27% usual care), a 38% relative reduction (adjusted RR 0.62, p=0.007). Adverse event dropouts fell from 16% to 9% (RR 0.59, p=0.04).

By 24 weeks, depression scores dropped to 7.1 (vs 9.2, mean difference -1.92, p<0.001) and anxiety to 4.6 (vs 5.8, -1.39, p=0.002). These gains persisted up to six months, suggesting sustained benefits.

• 40% lower overall discontinuation risk.
• Significant anxiety co-morbidity improvements.
• Potential NHS savings from reduced switchovers.

While limitations like non-blinding and missing data exist, the pragmatic design mirrors real-world use.NIHR coverage.

Voices from the Frontline: Researchers and Patients Speak

Prof Cipriani notes: "Mental health lags behind other fields... PETRUSHKA personalises from the outset." NIHR's Mike Lewis hails it as "cutting-edge digital tools for smarter care." Participant Henry Winchester shared: "It found a milder option... life-changing."

These perspectives highlight the tool's acceptability, vital for adoption in UK higher education-linked NHS services.

Implications for the NHS and UK Higher Education

With depression costing the UK economy £39 billion yearly, PETRUSHKA could cut non-adherence, freeing GP time. Oxford Health BRC's involvement underscores universities' role in translating research to practice. For academics, it opens doors in research jobs on AI ethics, implementation science, and mental health tech.

Precision Psychiatry: Oxford's Leadership and UK Peers

Prof Cipriani's lab pioneers evidence-based tools, building on prior meta-analyses. UK universities like Sheffield and UCL explore similar AI for depression triage. NIHR funds multiple initiatives, positioning Britain as a precision psychiatry hub.Explore AI in UK higher ed.

Challenges, Criticisms, and the Road Ahead

Critics note the trial's open-label design and attrition rates, urging blinded validation. Ethical concerns around AI 'black boxes' persist, though PETRUSHKA's transparency helps. Future phases target NHS rollout, with expansions to pharmacogenomics.

Stakeholders call for training; universities offer career advice for AI-psychiatry roles.

Photo by Artyom Korshunov on Unsplash

Career Opportunities in AI-Driven Mental Health Research

This breakthrough fuels demand for experts in computational psychiatry. UK universities seek postdocs, lecturers in precision medicine. Explore openings at higher-ed-jobs, university-jobs, or UK listings. Rate professors shaping the field via Rate My Professor.

Precision psychiatry isn't just treatment—it's a career frontier blending AI, neuroscience, and clinical care.