Home / Research / Literature Review

Literature Review

Annotated bibliography of key papers and meta-analyses in digital mental health, organized by topic. We've included the papers that inform our work, with brief annotations on their significance.

HRV Biofeedback

Physiological mechanisms and efficacy

Digital Engagement

Why users drop out and what helps

AI Safety in Health

Crisis response, bias, ethics

Digital Therapeutics

Evidence for digital interventions

Measurement-Based Care

Routine outcome monitoring

Systems Reform

IAPT, CCBHCs, stepped care

HRV Biofeedback and Breathing Interventions

Heart Rate Variability (HRV) biofeedback and resonance breathing represent some of the strongest evidence for digital physiological interventions.

Goessl, V.C., Curtiss, J.E., & Hofmann, S.G. (2017)

"The effect of heart rate variability biofeedback training on stress and anxiety: a meta-analysis"

Psychological Medicine, 47(15), 2578-2586.

Key Finding: Meta-analysis of 24 studies found HRV biofeedback significantly reduced self-reported stress and anxiety (Hedges' g = 0.81). Effect robust across different populations.

Why It Matters: Effect size of 0.81 is large—comparable to or exceeding pharmacological interventions. Establishes HRV biofeedback as evidence-based.

Lehrer, P.M., & Gevirtz, R. (2014)

"Heart rate variability biofeedback: how and why does it work?"

Frontiers in Psychology, 5, 756.

Key Finding: Comprehensive review of mechanisms. Resonance frequency breathing (~0.1 Hz / 6 breaths per minute) maximizes baroreflex gain and HRV amplitude.

Why It Matters: Explains the "why" behind the intervention. Clear mechanism supports confident recommendation.

Porges, S.W. (2007)

"The polyvagal perspective"

Biological Psychology, 74(2), 116-143.

Key Finding: Polyvagal Theory provides framework for understanding vagal nerve's role in social engagement and stress response. High vagal tone associated with resilience.

Why It Matters: Theoretical foundation for why HRV interventions affect emotional regulation. Connects physiology to social/emotional functioning.

Vaschillo, E., Vaschillo, B., & Lehrer, P. (2006)

"Characteristics of resonance in heart rate variability stimulated by biofeedback"

Applied Psychophysiology and Biofeedback, 31(2), 129-142.

Key Finding: Identified individual resonance frequencies (typically 0.075-0.12 Hz). Resonance breathing at individual's frequency maximizes HRV amplitude.

Why It Matters: Establishes ~6 breaths/min as typical optimal rate, though individual variation exists. Guides intervention design.

Digital Mental Health Engagement

Understanding why users abandon mental health apps—and what predicts sustained engagement—is critical for building effective tools.

Baumel, A., Muench, F., Edan, S., & Kane, J.M. (2019)

"Objective User Engagement With Mental Health Apps: Systematic Search and Panel-Based Usage Analysis"

Journal of Medical Internet Research, 21(9), e14567.

Key Finding: Analysis of actual usage data showed median 14-day retention of only 3.9% for mental health apps. Most users stop within days of download.

Why It Matters: The definitive paper on the "engagement crisis." Forces honest assessment of app efficacy claims vs. actual use patterns.

Torous, J., Nicholas, J., Larsen, M.E., Firth, J., & Christensen, H. (2018)

"Clinical review of user engagement with mental health smartphone apps: Evidence, theory and improvements"

Evidence-Based Mental Health, 21(3), 116-119.

Key Finding: Reviews engagement predictors: human support, personalization, gamification (with caveats), push notifications. Lack of engagement threatens all efficacy claims.

Why It Matters: Practical guidance from leading researcher on improving engagement. Emphasizes hybrid human-digital models.

Mohr, D.C., Weingardt, K.R., Reddy, M., & Schueller, S.M. (2017)

"Three Problems With Current Digital Mental Health Research... and Three Things We Can Do About Them"

Psychiatric Services, 68(5), 427-429.

Key Finding: Identifies three core problems: (1) efficacy-effectiveness gap, (2) not designing for implementation, (3) ignoring context. Proposes solutions.

Why It Matters: Essential critique of the field. Why RCT results don't translate to real-world impact.

AI Safety in Mental Health

The risks of AI in mental health contexts—from crisis detection failures to bias—are increasingly documented.

Torous, J., & Roberts, L.W. (2017)

"The Ethical Use of Mobile Health Technology in Clinical Psychiatry"

Journal of Nervous and Mental Disease, 205(1), 4-8.

Key Finding: Framework for ethical deployment of digital tools in psychiatry. Emphasizes autonomy, safety, data privacy, and maintaining therapeutic relationship.

Why It Matters: Foundational ethics framework from leading voices in digital psychiatry. Guides responsible development.

Miner, A.S., et al. (2016)

"Smartphone-Based Conversational Agents and Responses to Questions About Mental Health, Interpersonal Violence, and Physical Health"

JAMA Internal Medicine, 176(5), 619-625.

Key Finding: Tested Siri, Google Now, Cortana, S Voice on crisis statements. Inconsistent, often inappropriate responses to suicidal ideation, depression, domestic violence.

Why It Matters: Early documentation of AI crisis response failures. Still relevant; problems persist in current LLMs.

Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019)

"Dissecting racial bias in an algorithm used to manage the health of populations"

Science, 366(6464), 447-453.

Key Finding: Widely used healthcare algorithm exhibited racial bias, affecting ~200 million Americans. Black patients assigned same risk score as white patients were actually sicker.

Why It Matters: Landmark paper on algorithmic bias in healthcare. Demonstrates need for careful bias auditing in mental health AI.

Martinez-Martin, N., et al. (2020)

"Ethics of Digital Mental Health During COVID-19: Crisis and Opportunities"

JMIR Mental Health, 7(12), e23776.

Key Finding: COVID accelerated digital mental health adoption but raised ethical concerns about data practices, evidence base, equity, and crisis safety.

Why It Matters: Contemporary ethics analysis. Crisis context doesn't excuse ethical shortcuts.

Digital Therapeutics Evidence

Meta-analyses and systematic reviews of digital mental health interventions.

Cuijpers, P., et al. (2019)

"Internet-Based vs. Face-to-Face Cognitive Behavior Therapy for Psychiatric and Somatic Disorders: An Updated Systematic Review and Meta-Analysis"

Cognitive Behaviour Therapy, 48(1), 1-18.

Key Finding: Internet CBT (iCBT) shows equivalent effects to face-to-face CBT for depression and anxiety when therapist-guided. Unguided iCBT shows smaller effects.

Why It Matters: Evidence for digital therapy delivery—but human support remains important for best outcomes.

Firth, J., et al. (2017)

"The efficacy of smartphone-based mental health interventions for depressive symptoms: a meta-analysis of randomized controlled trials"

World Psychiatry, 16(3), 287-298.

Key Finding: Meta-analysis of 18 RCTs found smartphone interventions significantly reduced depressive symptoms (g = 0.38). Effect stronger for mild-moderate depression.

Why It Matters: Robust evidence for smartphone-delivered interventions, with appropriate caveats about severity.

Linardon, J. (2020)

"Can acceptance, mindfulness, and self-compassion be learned by smartphone apps? A systematic and meta-analytic review of RCTs"

Behavior Therapy, 51(4), 646-658.

Key Finding: App-based mindfulness interventions improve mindfulness skills (g = 0.55), self-compassion (g = 0.45), and reduce distress (g = 0.31).

Why It Matters: Evidence for mindfulness apps—though effects smaller than guided programs.

Measurement-Based Care

Evidence for routine outcome monitoring in mental health treatment.

Fortney, J.C., et al. (2017)

"A Tipping Point for Measurement-Based Care"

Psychiatric Services, 68(2), 179-188.

Key Finding: Reviews evidence for measurement-based care (MBC). MBC improves outcomes, but implementation barriers persist. Technology can facilitate adoption.

Why It Matters: Makes the case for routine measurement as standard of care. Digital tools can reduce burden.

Clark, D.M. (2018)

"Realizing the Mass Public Benefit of Evidence-Based Psychological Therapies: The IAPT Program"

Annual Review of Clinical Psychology, 14, 159-183.

Key Finding: Comprehensive review of IAPT by its architect. Routine outcome measurement is central to quality and accountability. 50%+ recovery rates at scale.

Why It Matters: Demonstrates population-scale measurement-based care is achievable and effective.

Systems Reform Literature

Archer, J., et al. (2012)

"Collaborative care for depression and anxiety problems"

Cochrane Database of Systematic Reviews, Issue 10.

Key Finding: Meta-analysis of 79 RCTs found collaborative care significantly improves depression and anxiety outcomes compared to usual care, with NNT of 5-7.

Why It Matters: Strong evidence for integrated care models. Basis for primary care mental health integration.

Patel, V., et al. (2018)

"The Lancet Commission on global mental health and sustainable development"

The Lancet, 392(10157), 1553-1598.

Key Finding: Comprehensive analysis of global mental health burden and solutions. Calls for task-shifting, integration, and scaled interventions.

Why It Matters: Authoritative global framework. Emphasizes systemic solutions over individual interventions.

SAMHSA (2020)

"National Guidelines for Behavioral Health Crisis Care: Best Practice Toolkit"

Substance Abuse and Mental Health Services Administration.

Key Finding: Evidence-based guidelines for crisis care continuum: crisis lines, mobile teams, stabilization units. Cost-effective alternatives to ED boarding.

Why It Matters: Implementation guide for Crisis Now model. Foundation for 988 system design.

Dr. John Torous: Selected Publications

Dr. Torous is one of the leading researchers in digital psychiatry. We highlight key papers from his extensive bibliography that inform our work.

Torous, J., et al. (2021)

"The growing field of digital psychiatry: current evidence and the future of apps, social media, chatbots, and virtual reality"

World Psychiatry, 20(3), 318-335.

Key Finding: Comprehensive review of digital psychiatry landscape. Cautious optimism balanced with honest assessment of evidence gaps and risks.

Torous, J., & Keshavan, M. (2018)

"A new window into psychosis: The rise digital phenotyping, smartphone assessment, and mobile monitoring"

Schizophrenia Research, 197, 67-68.

Key Finding: Introduces digital phenotyping concepts for psychosis research. Passive smartphone data can provide insights into illness patterns.

Torous, J., et al. (2020)

"Towards a consensus around standards for smartphone apps and digital mental health"

World Psychiatry, 19(1), 97-98.

Key Finding: Call for standards in digital mental health: evidence base, safety, data practices, usability. Framework for app evaluation.

Contradictory Evidence & Open Questions

We also document where evidence is mixed or where studies disagree. Intellectual honesty requires acknowledging uncertainty.

Areas of Genuine Uncertainty
  • Long-term effects of digital interventions: Most studies are short-term; durability unknown
  • Active ingredients: What specifically causes improvement in digital interventions?
  • Negative effects: Potential harms are understudied
  • Moderators: Who benefits most (and least) from digital approaches?
  • Implementation at scale: RCT results may not translate to real-world deployment

Publication Bias Concerns

Mental health app research may be subject to significant publication bias:

  • Industry-funded studies more likely to show positive results
  • Negative trials may go unpublished
  • Many apps have no independent evaluation
  • Systematic reviews may overestimate effects

We try to cite independent research where available and note industry funding when relevant.

← Research Home

Return to research overview.

Whitepaper →

Full synthesis of evidence and recommendations.