ScanSkinAI Research · Narrative review and conceptual framework

ScanSkinAI and Longitudinal Skin Self Care: A Narrative Evidence Review and Proposed Evaluation Framework

What should someone be able to understand and do after an AI skin screening result? This paper reviews the evidence for digital skin self-care and proposes how ScanSkinAI should connect a scan to useful, safe follow-up — and how that should be tested.

Paper information

Author
Dr Lifeng Zhu — Founder and Chief Executive Officer, Ivy AI Solutions Limited
Affiliation
Ivy AI Solutions Limited, United Kingdom and Hong Kong
Published
14 November 2026 · Version 1.0.0
Article type
Narrative review and conceptual framework

This is a conceptual article. It analyses no ScanSkinAI user data, has not been through journal peer review and does not show that ScanSkinAI improves health outcomes. ScanSkinAI does not provide a medical diagnosis — see a healthcare professional about any new, changing or worrying skin concern.

Quick answer

This narrative evidence review by Dr Lifeng Zhu (Ivy AI Solutions Limited, November 2026) asks how ScanSkinAI could connect a one-off AI skin screening result to safe, useful self-care over time. It reviews five selected eczema and acne trials, finds that structured digital self-management alongside usual care is plausible but modest and hard to attribute to single features, and proposes a framework that links a clear result, a chosen action, optional records and purposeful follow-up, with professional assessment always taking priority. It analyses no user data and does not show that ScanSkinAI improves health outcomes, engagement or subscription renewal; it sets out how those should be tested.

Key takeaways

  1. 1A care library is not the same as a longitudinal intervention: each guide also needs an action, a timescale, what a follow-up checks and when to stop self-monitoring and see a professional.
  2. 2Selected trials (Eczema Care Online, a 2026 caregiver education trial, an acne web intervention, the Atopic App and an acne text-reminder trial) support plausibility, but effects were modest and came alongside usual care.
  3. 3Reminders alone are not reliable: in a 40-patient acne trial, twice-daily texts did not improve adherence (33.9% vs 36.5%).
  4. 4Escalation is a valid outcome. A new or changing mole, or any high-risk finding, should go to professional assessment rather than wait for a future comparison photo.
  5. 5Patient benefit, meaningful engagement and subscription renewal are separate outcomes and must be measured separately.
  6. 6A staged plan is proposed: user and clinician co-design, then a feasibility study in adults with clinician-confirmed eczema (POEM at 12 weeks), then separate commercial experiments.

Abstract

Background
Image-based skin screening may leave users uncertain about what to do after receiving a result. For ScanSkinAI, a proposed dermatologist-authored care library creates an opportunity to connect screening with practical education and follow-up. However, evidence supporting digital self-management cannot be assumed to establish platform effectiveness or commercial retention.
Objective
To develop a ScanSkinAI-specific framework connecting educational guidance, user-selected actions and longitudinal feedback, and to specify how its clinical, engagement and commercial effects should be evaluated.
Methods
A targeted narrative review examined selected primary trials in eczema and acne, behavioural intervention frameworks and relevant methodological guidance. Two developer-provided ScanSkinAI planning documents and public platform descriptions were examined as design materials, not effectiveness evidence. Findings were mapped to proposed platform components and testable mechanisms. No individual-level user data were analysed.
Evidence synthesis
Selected studies support the plausibility of structured digital self-management alongside usual care, while showing important limitations in durability, generalisability and attribution to individual components. Engagement with core content can exceed uptake of optional modules, but these observations do not establish that a particular interface improves outcomes. No evidence reviewed establishes the effect of the proposed ScanSkinAI care library on subscription conversion or renewal.
Proposed framework
The framework connects understandable results with clinically governed education, a manageable chosen action, optional symptom and photo records, and purposeful follow-up. It separates appropriate clinical escalation from routine monitoring and distinguishes meaningful engagement from commercial retention. Public education and subscriber tools share a versioned clinical knowledge base.
Conclusions
ScanSkinAI provides a concrete setting for evaluating continuity between screening and self-care. Prospective work should establish feasibility, safety and patient benefit before interpreting higher engagement or renewal as evidence of success.

Keywords: ScanSkinAI; digital dermatology; self-management; patient education; artificial intelligence; meaningful engagement; longitudinal monitoring; subscription retention

Research questions

A directory of condition information supports access to knowledge. A longitudinal intervention must also specify an action, when it is suitable, what a later observation means and when self-monitoring should give way to professional assessment. The paper asks:

  1. Q1

    How should ScanSkinAI translate a screening result into an appropriate educational action?

  2. Q2

    What constitutes a useful follow-up for different user needs?

  3. Q3

    How can patient outcomes and meaningful engagement be distinguished from subscription behaviour?

  4. Q4

    What prospective design would test these relationships without treating commercial continuation as evidence of health benefit?

Methods

A targeted narrative review, not a systematic review. The analysis keeps three levels of inference apart: published findings from other interventions, proposed ScanSkinAI design choices, and questions that still need prospective testing.

  • Targeted web-based literature discovery and source verification on 14 November 2026, prioritising publisher pages, PubMed-indexed records and original methodological publications.
  • Five purposively selected trial reports in eczema and acne. This was not a systematic review: no PRISMA flow, meta-analysis, formal risk-of-bias score or certainty grade.
  • Two unpublished developer planning documents (the Skin Condition Care Library proposal and a UI/UX plan dated 12 November 2026) were treated as design materials, not evidence of effectiveness.
  • Findings were mapped to the capability–opportunity–motivation model of behaviour, person-based development and effective-engagement principles.
  • No customer interviews, cancellation surveys, clinical records, billing events or individual-level telemetry were analysed.

Evidence synthesis

The selected trials support the plausibility of structured digital self-care alongside usual care, but effects were modest, durability and generalisability were limited, and benefits could not be attributed to single components. None evaluated a ScanSkinAI paid offer, so no pooled effect was calculated.

Table 1. Selected clinical evidence and transfer boundaries
  • Eczema Care Online [9]

    Population and design
    Two pragmatic trials; 677 parents/carers and young people
    Finding relevant to ScanSkinAI
    Modest sustained symptom benefit
    Principal transfer boundary
    Adjunct to usual care; condition-specific populations
  • Digital caregiver education [10]

    Population and design
    Multicentre trial; 615 young children
    Finding relevant to ScanSkinAI
    Lower short-term relapse
    Principal transfer boundary
    Combined intervention; durability uncertain
  • Acne web intervention [11]

    Population and design
    Feasibility trial; 53 young people
    Finding relevant to ScanSkinAI
    High core-module completion
    Principal transfer boundary
    No randomised comparison of interface designs
  • Atopic App [12]

    Population and design
    Three-group trial; 58 children
    Finding relevant to ScanSkinAI
    Use-outcome associations
    Principal transfer boundary
    Engagement-based comparisons are confounded
  • Acne reminders [13]

    Population and design
    Trial; 40 patients
    Finding relevant to ScanSkinAI
    No significant adherence advantage
    Principal transfer boundary
    Small sample; reminder component only

What each study found

Eczema Care Online

Two pragmatic randomised trials (340 parents or carers of children; 337 young people). Compared with usual care, adjusted POEM differences over 24 weeks were −1.5 points (95% CI −2.5 to −0.6) and −1.9 points (−3.0 to −0.8), with benefits persisting to 52 weeks. Average effects were modest and came from behavioural support alongside existing care.

Smartphone caregiver education (2026)

615 children aged 0–6 years across 12 centres. Relapse at 12 weeks was 16.6% with the intervention vs 24.0% with control care (relative risk 0.69, 95% CI 0.50–0.96). Later relapse differences and secondary outcomes were not statistically significant. The package included action planning and clinician access, so the effect cannot be attributed to written content alone.

Atopic App

58 children in three groups. All groups improved; analyses linking more app use to better outcomes were vulnerable to self-selection, so they do not show that encouraging more use causes more improvement.

Acne web intervention

53 young people randomised. 23 of 26 in the intervention arm completed the core module, while individual optional modules were visited by 12%–42%. Users were not randomised to different navigation designs, so this cannot prove a core-first interface is superior.

Acne text reminders

40 patients randomised to twice-daily text messages or none during topical acne treatment. Electronically measured adherence was 33.9% vs 36.5% (P=.75), with similar clinical improvement — a caution against treating notifications as a reliable intervention on their own.

A proposed framework for ScanSkinAI

The programme theory: a result becomes more useful when it is connected to a suitable action and a purposeful chance to review that action. These are hypothesised pathways from behavioural theory (capability, opportunity, motivation), not established ScanSkinAI mechanisms.

The five-step pathway

  1. 1

    Interpret the result

    An understandable explanation of the screening result, its uncertainty and its limits.

  2. 2

    Confirm the concern and goal

    The user confirms what they are concerned about and what they want to achieve — including an “unsure” option.

  3. 3

    Choose a suitable action

    A modest educational action compatible with any existing clinical instructions; never an inferred prescription.

  4. 4

    Record when useful

    Optional symptom check-ins and photos, spaced to match the purpose of the follow-up.

  5. 5

    Review or seek care

    Review the record against the goal, or move to professional assessment. Escalation can happen at any step and overrides routine prompts.

Escalation first. A decision to seek professional assessment can happen at any step and always overrides routine engagement prompts.

Table 2. Proposed components and observable mechanisms
  • Result explanation

    Hypothesised mechanism
    Better understanding of uncertainty and next steps
    Observable indicator
    Comprehension of the result and its limits
    Boundary
    Explanation does not establish accuracy
  • User-confirmed concern

    Hypothesised mechanism
    Greater relevance and agency
    Observable indicator
    Appropriate concern and goal selection
    Boundary
    Selection does not confirm diagnosis
  • Practical educational action

    Hypothesised mechanism
    Reduced difficulty acting on guidance
    Observable indicator
    Ability to describe and complete the chosen task
    Boundary
    No unsupported prescribing
  • Optional symptom and photo record

    Hypothesised mechanism
    Better recall and reflection
    Observable indicator
    Completed relevant observations
    Boundary
    Photo differences may reflect capture conditions

Design principles

From guide template to usable action

Each guide should state its audience and exclusions, a plain-language explanation, general care principles, common mistakes, optional actions, realistic expectations, recording prompts, a purposeful follow-up, escalation advice and clinical provenance. A check-in should return something understandable — not just transfer logging work to the user.

Personalisation and uncertainty

Choosing an article for a followed concern is content selection, not personalised clinical management. The system should keep track of whether a label came from the user, a clinician or an AI suggestion, and keep cosmetic and clinical pathways separate — a cosmetic redness metric should not establish rosacea.

Timing and meaningful engagement

Return intervals should reflect purpose, burden and clinical context. NICE notes acne treatment benefits can take 6–8 weeks and recommends review at 12 weeks, so a short interval need not show improvement. Low use can be appropriate when a task is complete; reassurance-seeking and notification burden should be monitored.

Public education and subscriber functions

Foundational education and safety information stay public; membership supports extended modules, organisation and longitudinal tools, all drawing on one versioned clinical knowledge base. Payment should never decide whether someone can understand a result’s limits, retrieve a purchased report or reach referral advice.

Safety, equity and governance

  • Continuity can increase risk if it reinforces a wrong first label, so input-quality guidance, uncertainty explanations and escalation come first. New care features do not validate the upstream image model.
  • A new or changing mole should be professionally assessed (NHS guidance); users should not be encouraged to postpone assessment until a future comparison.
  • High-risk or unresolved findings should not enter an ordinary routine-building sequence because it is commercially desirable.
  • Clinical content needs a named qualified author, documented sources, a version and a review schedule. A reviewer credit must not be presented as authorship or as individual patient care.
  • Equity must be evaluated: dermatology AI has shown performance gaps across skin tones and uncommon conditions in external benchmarks, and access, language, literacy, device quality and affordability all matter.
  • Linked condition records are sensitive health data (ICO). Collection should be minimised, access restricted, retention defined and clinical records kept separate from routine marketing systems.
  • Intended purpose and claims should be assessed against the applicable regulatory framework (MHRA). The paper makes no determination about ScanSkinAI’s regulatory classification.

Worried about a mole that is new or changing? Read when monitoring is suitable or ask for a dermatologist review.

Prospective evaluation agenda

The paper proposes learning in stages before expanding the catalogue, and keeps clinical, engagement and commercial questions in separate studies so that higher use or renewal is never read as proof of patient benefit.

  1. Stage 1

    Formative development

    Iterative interviews and usability sessions with intended users and qualified dermatology professionals, deliberately sampling different literacy levels, devices, skin tones and reasons for use. Tasks test understanding of uncertainty, educational vs prescribed care, choosing an action, symptom-only check-ins and recognising escalation advice.

  2. Stage 2

    Feasibility and patient outcomes

    A first feasibility study in adults with clinician-confirmed atopic dermatitis receiving usual care, then a randomised comparison of usual care plus static education vs usual care plus guided education, chosen actions and structured follow-up. Candidate primary endpoint: baseline-adjusted POEM at 12 weeks, with DLQI for quality of life. Sample size follows the clinical estimand, never an assumed conversion uplift.

  3. Stage 3

    Engagement and commercial evaluation

    Separate randomised product evaluations. Non-subscribers: existing offer vs a useful care preview, with price, billing terms and core scan access unchanged (first subscription payment within 7 days). Subscribers: successful renewal at the first eligible billing date, distinguishing voluntary cancellation from failed payment.

Table 3. Proposed outcome domains and interpretation
  • Clinical

    Candidate outcome
    Baseline-adjusted POEM in an eczema trial
    Timing
    12 weeks and later follow-up
    Interpretation
    Patient symptom outcome
  • Quality of life

    Candidate outcome
    Suitable validated dermatology measure
    Timing
    Baseline and follow-up
    Interpretation
    Patient impact, not platform usage
  • Understanding

    Candidate outcome
    Comprehension of limits and next steps
    Timing
    First session and early follow-up
    Interpretation
    Intervention-specific measure needing validation
  • Engagement

    Candidate outcome
    Meaningful action during days 22–35
    Timing
    Approximately one month
    Interpretation
    Process indicator, not a clinical surrogate

Analysis transparency and independence

  • Allocation concealed until assignment; analysts or assessors blinded where feasible; effect sizes reported with uncertainty intervals.
  • Missing outcomes and non-use reported separately — neither assumed to mean improvement or failure.
  • Protocol finalised with qualified investigators, ethics review and prospective registration where required (SPIRIT 2025; CONSORT 2025 and CONSORT-EHEALTH for reporting).
  • Independent clinical oversight, an agreed statistical analysis plan and publication of results regardless of direction, including results favouring referral, lower use or non-renewal.

What this paper does not show

  • That ScanSkinAI or My Skin Care improves skin symptoms or quality of life
  • ScanSkinAI’s diagnostic accuracy or the performance of its image model
  • Any effect on engagement, subscription conversion or renewal
  • Results from ScanSkinAI users — no individual-level user data were analysed
  • A systematic or exhaustive review of all relevant studies
  • A registered trial protocol, ethics approval or regulatory classification

Limitations

  • Purposive, non-systematic review with uneven full-text access, no independent duplicate screening and no formal risk-of-bias assessment; eczema and acne evidence dominates.
  • The documentary analysis relies on developer-supplied plans and public descriptions and does not establish which functions are operational or how users experience them.
  • The framework has not undergone external consensus development or user testing; the evaluation is an agenda, not a registered protocol.
  • The sole named author leads the company that develops ScanSkinAI — a material competing interest that disclosure cannot eliminate.

Conclusion

ScanSkinAI offers a specific setting for studying how image-based screening can connect to longitudinal educational support. The central requirement is a useful relationship between guidance, a suitable user action and an interpretable follow-up, with clear routes to professional assessment. A larger library alone does not establish that relationship. The effectiveness of the ScanSkinAI care experience remains to be established.

These pages describe tools and evidence mentioned in or related to the paper. The paper does not evaluate their effectiveness.

Declarations

Competing interests
Dr Lifeng Zhu is the founder and chief executive officer of Ivy AI Solutions Limited, the company developing ScanSkinAI. The framework concerns the company’s product and therefore presents a direct commercial competing interest.
Ethics and consent
A literature-based synthesis and analysis of platform design documents. No participant recruitment, intervention delivery or individual-level health-data analysis is reported. Any future human-participant study requires prior ethics determination, consent arrangements and prospective registration where applicable.
Data availability
No new participant dataset was generated. The developer-provided planning documents are unpublished design materials; their relevant characteristics are described in the paper.
Patient and public involvement
No patient or public involvement activity was conducted for this conceptual article. Such involvement is proposed for subsequent formative development.
AI assistance
ChatGPT, an OpenAI language-model system, assisted with literature-source discovery, drafting and organisation. It did not analyse participant data. The human author remains responsible for accuracy, attribution and integrity.

Correspondence: info@scanskinai.com

References

  1. 1.Ivy AI Solutions Limited. ScanSkinAI [Internet]. Accessed 14 November 2026. View page
  2. 2.Ivy AI Solutions Limited. Chronic skin condition and mole tracker [Internet]. Accessed 14 November 2026. View page
  3. 3.Ivy AI Solutions Limited. ScanSkinAI Skin Condition Care Library. Unpublished developer-provided planning document; version supplied for review November 2026.
  4. 4.Ivy AI Solutions Limited. ScanSkinAI UI/UX Improvement Plan to Reduce Subscription Cancellations. Version 1.0. 12 November 2026. Unpublished developer-provided planning document.
  5. 5.Skivington K, Matthews L, Simpson SA, et al. A new framework for developing and evaluating complex interventions: update of Medical Research Council guidance. BMJ. 2021;374:n2061. https://doi.org/10.1136/bmj.n2061
  6. 6.Yardley L, Morrison L, Bradbury K, Muller I. The person-based approach to intervention development: application to digital health-related behavior change interventions. J Med Internet Res. 2015;17(1):e30. https://doi.org/10.2196/jmir.4055
  7. 7.Yardley L, Spring BJ, Riper H, et al. Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med. 2016;51(5):833-842. https://doi.org/10.1016/j.amepre.2016.06.015
  8. 8.Eysenbach G. The law of attrition. J Med Internet Res. 2005;7(1):e11. https://doi.org/10.2196/jmir.7.1.e11
  9. 9.Santer M, Muller I, Becque T, et al. Eczema Care Online behavioural interventions to support self-care for children and young people: two independent, pragmatic, randomised controlled trials. BMJ. 2022;379:e072007. https://doi.org/10.1136/bmj-2022-072007
  10. 10.Yang H, Shu H, Wang LH, et al. Smartphone-based digital eczema education program for atopic dermatitis in children aged 0 to 6 years: multicenter, randomized, parallel controlled clinical study. J Med Internet Res. 2026;28:e79559. https://doi.org/10.2196/79559
  11. 11.Ip A, Muller I, Geraghty AWA, et al. Supporting self-management among young people with acne vulgaris through a web-based behavioral intervention: development and feasibility randomized controlled trial. JMIR Dermatol. 2021;4(2):e25918. https://doi.org/10.2196/25918
  12. 12.Zvulunov A, Lenevich S, Migacheva N. Mobile health app as an auxiliary tool in management of atopic dermatitis in children: randomized controlled trial. JMIR Dermatol. 2025;8:e60479. https://doi.org/10.2196/60479
  13. 13.Boker A, Feetham HJ, Armstrong A, Purcell P, Jacobe H. Do automated text messages increase adherence to acne therapy? Results of a randomized, controlled trial. J Am Acad Dermatol. 2012;67(6):1136-1142. https://doi.org/10.1016/j.jaad.2012.02.031
  14. 14.Michie S, van Stralen MM, West R. The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci. 2011;6:42. https://doi.org/10.1186/1748-5908-6-42
  15. 15.National Institute for Health and Care Excellence. Acne vulgaris: management. NICE guideline NG198. Accessed 14 November 2026. View source
  16. 16.NHS. Moles [Internet]. Accessed 14 November 2026. View source
  17. 17.Daneshjou R, Vodrahalli K, Novoa RA, et al. Disparities in dermatology AI performance on a diverse, curated clinical image set. Sci Adv. 2022;8:eabq6147. https://doi.org/10.1126/sciadv.abq6147
  18. 18.Information Commissioner’s Office. What is special category data? Accessed 14 November 2026. View source
  19. 19.Medicines and Healthcare products Regulatory Agency. Crafting an intended purpose in the context of software as a medical device. 2023. Accessed 14 November 2026. View source
  20. 20.Charman CR, Venn AJ, Williams HC. The patient-oriented eczema measure: development and initial validation of a new tool for measuring atopic eczema severity from the patients’ perspective. Arch Dermatol. 2004;140(12):1513-1519. https://doi.org/10.1001/archderm.140.12.1513
  21. 21.Finlay AY, Khan GK. Dermatology Life Quality Index (DLQI): a simple practical measure for routine clinical use. Clin Exp Dermatol. 1994;19(3):210-216. https://doi.org/10.1111/j.1365-2230.1994.tb01167.x
  22. 22.Chan AW, Boutron I, Hopewell S, et al. SPIRIT 2025 statement: updated guideline for protocols of randomised trials. BMJ. 2025;389:e081477. https://doi.org/10.1136/bmj-2024-081477
  23. 23.Hopewell S, Chan AW, Collins GS, et al. CONSORT 2025 statement: updated guideline for reporting randomised trials. BMJ. 2025;389:e081123. https://doi.org/10.1136/bmj-2024-081123
  24. 24.Eysenbach G; CONSORT-EHEALTH Group. CONSORT-EHEALTH: improving and standardizing evaluation reports of web-based and mobile health interventions. J Med Internet Res. 2011;13(4):e126. https://doi.org/10.2196/jmir.1923
  25. 25.International Committee of Medical Journal Editors. Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. Updated January 2026. Accessed 14 November 2026. View source

Cite this paper

Zhu, L. (2026). ScanSkinAI and Longitudinal Skin Self Care: A Narrative Evidence Review and Proposed Evaluation Framework (Version 1.0.0). Zenodo. https://doi.org/10.5281/zenodo.24215072

Published under Creative Commons Attribution 4.0 International (CC BY 4.0). Reuse is permitted with attribution — please cite the DOI.

Frequently asked questions

It is a narrative evidence review and proposed evaluation framework by Dr Lifeng Zhu, published on Zenodo on 14 November 2026 (DOI 10.5281/zenodo.24215072). It asks how an AI skin screening result can be connected to safe, useful self-care and follow-up over time, and how the clinical, engagement and commercial effects of doing so should be tested.

No. The paper analyses no ScanSkinAI user data and reports no observed effect on clinical outcomes, engagement, subscription conversion or renewal. It reviews evidence from other digital self-care interventions and proposes how ScanSkinAI’s care features should be evaluated.

Five purposively selected trials in eczema and acne: Eczema Care Online, a 2026 smartphone caregiver education trial, an acne web intervention, the Atopic App and an acne text-reminder trial, plus behavioural and methodological guidance such as MRC complex-intervention guidance, the person-based approach and the behaviour change wheel.

Not reliably on their own. In the reviewed 40-patient acne trial, twice-daily text reminders did not improve electronically measured adherence (33.9% vs 36.5%, P=.75). The paper cautions against treating notifications as an independent intervention.

The framework treats professional assessment as a valid outcome at any step. A new or changing mole, or any high-risk or unresolved finding, should be assessed by a healthcare professional rather than waiting for a future comparison photo.

It is a company-authored manuscript archived on Zenodo with a permanent DOI. It has not been through independent journal peer review, and the author declares a direct commercial competing interest as founder and CEO of Ivy AI Solutions Limited.

Zhu, L. (2026). ScanSkinAI and Longitudinal Skin Self Care: A Narrative Evidence Review and Proposed Evaluation Framework (Version 1.0.0). Zenodo. https://doi.org/10.5281/zenodo.24215072

The Zenodo record is published under a Creative Commons Attribution 4.0 International (CC BY 4.0) licence, which allows reuse with attribution. Please cite the DOI.

This page summarises the archived manuscript for research transparency and general education. The version of record is the Zenodo publication. ScanSkinAI supports preliminary skin awareness and screening; it does not diagnose disease, rule out disease or replace a doctor or dermatologist.

Back to Research & Evidence