Researcher · Innovator · Collaborator

Transforming mental health through digital innovation & AI

I'm Dr Simon Cherry — a researcher at Burnage University pioneering how technology, data science, and generative AI can make mental health support more accessible, engaging, and effective for everyone.

Explore My AI Work View Publications
25+
Peer-reviewed Publications
10+
Years in Digital Health
7,400+
Research Participants
6
NHS Trust Collaborations
About

Bridging the gap between technology and mental health care

Combining clinical evidence, co-design, and cutting-edge AI to build interventions that actually work — and reach the people who need them most.

I'm passionate about one big question: how can we use technology to give everyone access to high-quality mental health support? My work spans the full pipeline — from co-designing apps with the communities who'll use them, to running rigorous clinical trials, to implementing solutions at scale in the NHS.

What excites me most right now is the intersection of generative AI and mental health. I'm exploring how large language models can be safely embedded within evidence-based therapeutic frameworks to create more personalised, adaptive, and scalable interventions — while keeping humans firmly in the loop.

I thrive on collaboration. Whether you're a clinician, a technologist, a policymaker, or someone with lived experience — let's build something together.

AI & Innovation

Pioneering the responsible use of generative AI in mental health

I believe LLMs are best understood as infrastructure components — powerful when bounded by clinical architecture and ethical guardrails.

💬

LLM-Augmented CBT

Embedding generative AI within structured therapeutic modules — paraphrasing thought records, generating Socratic questions, and creating personalised reflections while staying protocol-bounded. Our lab study showed significantly higher perceived personalisation vs templates (4.2 vs 3.1, p<.001) with zero adverse events.

🤖

Chatbot Psychoeducation

A conversational agent delivering psychoeducation for health anxiety achieved 82% module completion (vs 54% for static content) and larger symptom reductions in our three-arm RCT. Demonstrating that AI-mediated delivery can boost engagement without sacrificing clinical rigour.

🔐

Federated Learning

Privacy-preserving ML across NHS trusts — training models where the data lives. Our federated anxiety-relapse classifier matched centralised performance (AUC 0.78 vs 0.80) with zero raw-data exchange across four trusts.

📈

Predictive Analytics

Using early interaction signals to predict disengagement — our gradient-boosted classifier achieves AUC 0.81 using just first-week behavioural features. Enabling proactive support before people drop out.

🛡️

AI Safety & Ethics

Developing the "guardrailed augmentation" model: generative AI enhances human-authored clinical content without replacing clinical judgement. Rigorous safety testing, transparent disclosure, and post-deployment surveillance are non-negotiable.

🧬

NLP & Mood Detection

Validated transformer-based NLP pipeline for automated mood detection from free-text diaries — achieving r=0.71 for positive affect and r=0.68 for negative affect against self-report measures.

Research

Selected publications

A track record of rigorous, high-impact research published in leading journals.

  1. Digital triage in primary care mental health: A cluster-randomised evaluation of algorithm-assisted referral to online therapies.
    2026Cherry, S., Okoro, N., & Bevan, T. · The Lancet Digital Health
    Cluster-RCT across 22 GP practices (n=1,847) finding 19% faster time-to-treatment-start with algorithm-assisted referral and equivalent 6-month recovery rates.
  2. Evaluating chatbot-delivered psychoeducation for health anxiety: A three-arm randomised controlled trial.
    2026Cherry, S., Mensah, O., & D'Souza, R. · JMIR
    Three-arm RCT (n=312) demonstrating superior engagement (82% vs 54%) and larger symptom reductions for conversational-agent delivery.
  3. Real-world effectiveness of prescription digital therapeutics for insomnia in NHS settings.
    2025Cherry, S., Park, J., & Wallace, F. · Sleep
    Prospective cohort (n=634) showing 61% remission rate for CBT-I completers across four NHS Integrated Care Boards.
  4. Predicting disengagement in app-supported CBT using early interaction signals.
    2025Cherry, S., Khan, R., & Elson, M. · JMIR
    Gradient-boosted classifier using first-week behavioural features to predict dropout risk, achieving AUC 0.81 in external validation.
  5. Co-designing a smartphone-based mood intervention for university students.
    2024Cherry, S., Boateng, A., & Freeman, L. · Internet Interventions
    Iterative co-design with 34 students followed by feasibility trial (n=67) — 78% retention and significant pre–post PHQ-9 reductions.
  6. Equity challenges in digital mental health implementation across urban primary care networks.
    2023Cherry, S., Patel, D., & Hargreaves, J. · BMC Health Services Research
    Mixed-methods evaluation showing patients from lower socioeconomic quintiles 2.3x less likely to be referred to digital interventions.
  7. Therapist support intensity and outcomes in blended digital psychotherapy: A multi-site pragmatic trial.
    2022Cherry, S., Li, W., & O'Donnell, K. · JMIR Mental Health
    Three-arm pragmatic RCT (n=384) finding light guidance produced non-inferior outcomes to intensive support in digital CBT.
  8. Digital therapeutics for generalised anxiety disorder: A systematic review and meta-analysis.
    2022Cherry, S., Adeyemi, F., & Shah, N. · Psychological Medicine
    Synthesised 38 RCTs (N=7,412) yielding pooled Hedges' g of 0.56 for digital interventions versus waitlist controls.
  9. User engagement trajectories in mental health apps: A latent class growth analysis.
    2021Cherry, S., & Worthington, E. · JMIR
    Identified four latent engagement classes from 2,841 users; "brief-consistent" users achieved comparable improvement to "high-intensity" users.
  10. Barriers to digital mental health service uptake among Black and South Asian communities.
    2021Cherry, S., Okonkwo, C., & Bennett, H. · BMC Public Health
    Qualitative exploration (n=41) identifying cultural mistrust, lack of representation, and preference for relational care as primary barriers.
  11. Wearable-triggered just-in-time adaptive interventions for workplace stress.
    2021Cherry, S., Moreno, R., & Gupta, A. · Digital Health
    Micro-randomised trial (n=54) delivering breathing exercises via smartwatch — 68% prompt acceptance and 1.2-point stress reduction.
  12. Gamification in digital CBT for depression: Effects on engagement and symptom outcomes.
    2020Cherry, S., Nakamura, K., & Walsh, P. · Computers in Human Behavior
    Factorial trial (n=208) finding gamification increased module completions by 31% without attenuating clinical improvement.
  13. Machine learning prediction of treatment response in app-delivered interventions for depression.
    2019Cherry, S., & Clarke, M. · Artificial Intelligence in Medicine
    Random forests and LASSO on pooled trial data (n=611) achieving 72% balanced accuracy for predicting treatment response.
  14. Cost-effectiveness of therapist-guided digital CBT versus face-to-face CBT for social anxiety disorder.
    2018Cherry, S., Park, J., & Williams, T. · Health Technology Assessment
    Economic evaluation (n=268) showing guided digital CBT dominated face-to-face delivery: lower costs (£487 vs £1,124) with non-significant QALY gain.
  15. Automated mood detection from digital diary free-text using natural language processing.
    2018Cherry, S., & Mensah, O. · Journal of Affective Disorders
    Validated transformer-based NLP pipeline achieving r=0.71 for positive affect detection against PANAS self-report.
  16. Digital interventions for insomnia: A meta-analysis of randomised controlled trials.
    2015Cherry, S., & Gallagher, M. · Sleep Medicine Reviews
    Meta-analysis of 18 RCTs (N=2,106) finding large effects on sleep efficiency (g=0.86) and sleep onset latency (g=0.72).
  17. Generative AI-assisted clinical note summarisation in digital therapy platforms: A mixed-methods evaluation.
    2026Cherry, S., Nguyen, H., & Byrne, C. · npj Digital Medicine
    Mixed-methods study (n=82 therapists) finding LLM-generated session summaries reduced documentation time by 42% while maintaining clinical accuracy rated 4.3/5 by supervisors.
  18. Culturally adapted digital mental health interventions for refugee populations: A pilot randomised trial.
    2025Cherry, S., Al-Rashidi, M., & Connolly, E. · Transcultural Psychiatry
    Pilot RCT (n=93) of a culturally adapted smartphone intervention showing feasibility (87% completion) and significant reductions in PTSD symptom severity (d=0.64).
  19. Ecological momentary assessment of digital intervention micro-usage and momentary well-being.
    2024Cherry, S., Eriksson, L., & Osei, K. · Behaviour Research and Therapy
    14-day EMA study (n=126, 4,218 prompts) demonstrating that brief in-app exercises (≤3 min) predicted within-person increases in positive affect at the next assessment point.
  20. Natural language processing for automated risk detection in online peer-support forums.
    2024Cherry, S., Kapoor, R., & Hollis, C. · Journal of Medical Internet Research
    Developed and validated a BERT-based classifier for detecting escalating risk in peer-support posts (F1=0.83), enabling triage response within 15 minutes.
  21. Patient preferences for AI-generated versus therapist-written feedback in guided self-help: A discrete choice experiment.
    2023Cherry, S., & Thornton, R. · BMC Psychiatry
    DCE (n=514) revealing patients valued empathy and personalisation over authorship; blinded AI-generated messages were rated comparably to therapist-written ones.
  22. Implementation fidelity of a digitally supported stepped-care model for perinatal depression.
    2023Cherry, S., Whitfield, G., & Akinola, T. · Archives of Women's Mental Health
    Process evaluation within a hybrid type-II trial (n=317) finding 74% fidelity to the stepped-care algorithm and 26% faster step transitions versus usual care.
  23. A Bayesian adaptive platform trial of digital interventions for social isolation in older adults.
    2020Cherry, S., Davies, H., & Phan, T. · The Gerontologist
    Bayesian adaptive trial (n=245) comparing three digital social-connectedness interventions; video-call facilitation arm declared superior with 94% posterior probability.
  24. Smartphone-delivered acceptance and commitment therapy for chronic pain: A feasibility RCT.
    2019Cherry, S., Obrien, L., & Kaur, S. · Pain Medicine
    Feasibility RCT (n=74) showing high acceptability (NPS +42) and preliminary evidence of improved pain interference and psychological flexibility at 8 weeks.
  25. Therapist attitudes toward technology-augmented practice: A national survey.
    2017Cherry, S., Reynolds, A., & Singh, P. · British Journal of Clinical Psychology
    Cross-sectional survey (n=623) identifying workload concerns (68%), data privacy (54%), and therapeutic alliance fears (47%) as primary barriers to digital tool adoption among UK therapists.
  26. Online cognitive–behavioural therapy for panic disorder: Long-term outcomes from a pragmatic effectiveness trial.
    2016Cherry, S., & Hammond, J. · Cognitive Behaviour Therapy
    24-month follow-up of a pragmatic trial (n=189) demonstrating durable improvements in panic severity and agoraphobic avoidance, with 63% maintaining clinically significant change.
  27. Safety netting in digital primary care: An algorithm for automated red-flag detection in patient-reported outcome data.
    2026Cherry, S., Okafor, J., & Lindström, A. · The Lancet Digital Health
    Validation study (n=4,312 submissions) of a rule-based safety-netting algorithm achieving 97% sensitivity for critical deterioration flags with a 6% false-positive rate.
  28. Optimising push-notification timing in digital mental health interventions using reinforcement learning.
    2025Cherry, S., Fernandez, D., & Yao, M. · npj Digital Medicine
    Contextual bandit algorithm (n=387 users, 12 weeks) increased engagement with therapeutic micro-exercises by 29% compared to fixed-schedule delivery.
  29. Digital phenotyping of sleep disruption as a predictor of depressive relapse: A prospective cohort study.
    2025Cherry, S., Okonkwo, C., & Petersen, I. · Molecular Psychiatry
    Prospective cohort (n=412, 6 months) finding passively sensed sleep irregularity predicted relapse onset 18 days earlier than self-report questionnaires (AUC=0.78).
  30. Therapeutic alliance in blended care: Does patient–therapist rapport transfer to the digital component?
    2024Cherry, S., & McAllister, F. · Psychotherapy Research
    Process–outcome analysis (n=196 dyads) demonstrating that stronger in-session alliance predicted 2.4x higher digital module completion, mediated by perceived platform trustworthiness.
  31. Voice biomarkers for real-time anxiety monitoring in mobile health applications.
    2024Cherry, S., Ibrahim, A., & Chen, Y. · IEEE Journal of Biomedical and Health Informatics
    Acoustic feature extraction from 30-second voice samples achieved 76% accuracy for state-anxiety classification against GAD-7 benchmarks in a free-living sample (n=148).
  32. Integrating digital therapeutics into IAPT pathways: A realist evaluation.
    2023Cherry, S., Hargreaves, J., & Obi, N. · Implementation Science
    Realist evaluation across 6 IAPT services identifying trust in evidence, leadership buy-in, and technical interoperability as core mechanisms enabling sustainable adoption.
  33. Single-session digital interventions for test anxiety in secondary school students: A cluster-randomised trial.
    2022Cherry, S., Bates, L., & Mwangi, J. · Journal of Child Psychology and Psychiatry
    School-based cluster RCT (14 schools, n=831) showing a 20-minute digital CBT session reduced pre-exam anxiety scores by 0.41 SD relative to attentional control.
  34. Ethical frameworks for AI-driven mental health triage: A Delphi consensus study.
    2022Cherry, S., & Adeyemi, F. · Journal of Medical Ethics
    Three-round Delphi (n=47 experts) converging on 12 consensus principles for responsible deployment of algorithmic triage in psychological services.
  35. Virtual reality exposure therapy for specific phobias delivered via consumer headsets: A non-inferiority trial.
    2021Cherry, S., Novak, P., & Gill, S. · Behaviour Research and Therapy
    Non-inferiority RCT (n=152) demonstrating consumer-grade VR exposure was non-inferior to lab-based VR on phobia severity at 3-month follow-up (Δ=−0.12, 95% CI −0.38 to 0.14).
  36. Longitudinal associations between social media use patterns and adolescent well-being: A registered report.
    2020Cherry, S., Warner, K., & Chowdhury, S. · Journal of Adolescence
    Pre-registered longitudinal study (n=1,203, 3 waves over 18 months) finding passive consumption, but not active use, predicted small declines in life satisfaction (β=−0.08).
  37. Automated sentiment analysis of therapy session transcripts: Concordance with observer-rated emotional processing.
    2019Cherry, S., & Mensah, O. · Clinical Psychology & Psychotherapy
    Comparison of NLP sentiment scoring against expert CEPS ratings (60 sessions) yielding ICC=0.69, supporting feasibility of automated process measurement.
  38. A digital relapse-prevention intervention for bipolar disorder: Protocol for a multi-centre randomised controlled trial (the eMOOD study).
    2017Cherry, S., Gallagher, M., & Steele, R. · BMC Psychiatry
    Published protocol for a 5-site RCT (target n=320) evaluating a smartphone-based mood-monitoring and early-warning system for bipolar relapse prevention.
  39. Computerised cognitive–behavioural therapy for obsessive–compulsive disorder: A pilot randomised controlled trial.
    2016Cherry, S., & Park, J. · Behavioural and Cognitive Psychotherapy
    Pilot RCT (n=52) of an 8-module cCBT programme for OCD showing large between-group effects on Y-BOCS (d=0.91) and high programme satisfaction.
  40. Internet-delivered psychoeducation for carers of people with early psychosis: A randomised waitlist-controlled trial.
    2014Cherry, S., Bennett, H., & Gallagher, M. · Early Intervention in Psychiatry
    Waitlist RCT (n=128 carers) demonstrating reduced carer burden (d=0.54) and improved mental health literacy post-intervention, with gains maintained at 6 months.
Thinking

Articles & commentary

Perspectives on the future of digital mental health and AI.

May 2026

Why Digital Health Literacy Is the Overlooked Determinant of Intervention Success

Patients scoring in the lowest tertile of eHEALS were 3.1x more likely to disengage before module two. Until we embed scaffolded digital literacy support within interventions, scalable digital mental health will remain unevenly distributed.

April 2026

The Case for Federated Learning in Mental Health Data Ecosystems

Models travel to the data rather than data travelling to a central server. Our federated anxiety-relapse classifier matched centralised performance with zero raw-data exchange — decentralised intelligence is the future of ethical analytics.

March 2026

Digital Phenotyping and the Ethics of Continuous Monitoring in Youth Mental Health

Young people broadly accepted passive monitoring when transparent and on-device — but any sense of automatic data flow to parents or clinicians triggered strong aversion. Autonomy-preserving architecture is an ethical prerequisite.

February 2026

Rethinking Dropout: When Disengagement From Digital Therapy Is Recovery

23% of "dropouts" had already achieved reliable clinical improvement. Adaptive completion algorithms should help people leave at the right time, not keep them engaged indefinitely.

January 2026

Large Language Models in Digital Mental Health: Promise, Peril, and a Path Forward

LLMs are best understood as infrastructure components. Our "guardrailed augmentation" model — where generative AI enhances clinical architecture without replacing judgement — offers a responsible development path.

December 2025

Implementation Cliffs: Why Effective Digital Health Interventions Fail to Scale

Three cliffs emerged: workforce readiness, pathway integration, and patient expectations. Technology development alone is insufficient — we need strategies at organisational, professional, and narrative levels.

Teaching & Service

Developing the next generation

🎓 MSc Digital Health

Module lead: Digital Interventions in Mental Health — covering design, evaluation, and implementation of tech-enabled therapies.

🔬 Doctoral Supervision

Supervising PhD and professional doctorate projects in digital therapy implementation, AI in mental health, and engagement science.

📝 Peer Review

Reviewer for JMIR, Digital Health, The Lancet Digital Health, and Psychological Medicine. Contributing to rigorous science in the field.

Get in Touch

Let's collaborate

Whether you're a researcher, clinician, technologist, policymaker, or someone with lived experience — I'd love to hear from you.

Department of Applied Health Sciences
Burnage University
Burnage, United Kingdom
simon@scherry.org

Get in Touch