Agentic education · Learning science

Actionable companion · Accessibility & inclusion

Learning with agency: Artificial-intelligence–based developmental screening for pre-school children (ages 3–6): a scoping review of deployed tools, clinical validity, equity, and implementation in guidance-center settings

AI-assisted editorial implementation guide, separate from the original research. Proposed actions are not claims or recommendations attributed to the source authors. No independent peer review of this guide is asserted.

Read original paperSearch paper agentCC BY 4.0

Evidence and context

Title and abstract screening; stage labels describe reported settings, not verified numerical ages. Design labels require full-text appraisal. Reported stage(s): Early childhood · preschool, Children · stage unspecified. Health-professions learning must be distinguished from K–12 instruction.

Original authors’ abstract, reproduced under the linked source license:

Abstract Background Early identification of neurodevelopmental conditions during the pre-school years (ages 3–6) is critical for timely intervention, yet traditional questionnaire-based screeners suffer from low positive predictive value, late detection, and inequitable access. Artificial intelligence (AI) and machine-learning (ML) methods promise more objective, scalable identification, and two devices (Cognoa Canvas Dx, 2021; EarliPoint Evaluation, 2023) are already US Food and Drug Administration (FDA)-authorized. Recent reviews summarize the technical literature, but few address how deployed tools perform on clinical validity, equity, and real-world implementation, particularly outside high-income, English-speaking settings. Methods Following Joanna Briggs Institute methodology and the PRISMA extension for scoping reviews (PRISMA-ScR), we searched seven databases (1 January 2020–15 June 2026) and charted primary validation studies of AI/ML developmental-screening tools for children aged 72 months or younger. We mapped tools by modality, diagnostic performance, regulatory status, geographic origin, and, as a structured analytic axis, demographic subgroup (equity) reporting and deployment readiness. Results Sixteen primary AI-tool studies met inclusion, spanning eight modality classes (eye-tracking, multimodal questionnaire-plus-video, computer-vision home/clinical video, tablet/smartphone digital phenotyping, voice/acoustic ML, electronic-health-record risk prediction, questionnaire-based ML, and wearable sensors) and eight countries (USA, South Korea, Australia, Switzerland, France, Israel, China, Vietnam). Reported sensitivities ranged from approximately 75 to 99.2% and specificities from 78.9 to 95.5%. Only one tool demonstrated stable performance across sex, race, and ethnicity; subgroup-stratified validation was otherwise near-absent. No CE-marked autism-specific screening device was identified. Conclusions The evidence base has broadened in modality and geography but remains autism-concentrated and equity-blind, with a wide gap between analytic accuracy and primary-care deployment. We propose equity validation, deployment-readiness reporting, and an implementation pathway for low-resource guidance-center settings as priorities. Keywords: algorithmic equity, artificial intelligence, autism spectrum disorder, developmental screening, digital phenotyping, implementation science, machine learning, pre-school children

What kind of evidence is this?

Supporting AI learning science · Evidence synthesis

Stage: Early childhood · preschool; Children · stage unspecified. Screening labels describe reported context and study-design language. They are not a risk-of-bias rating or proof of effectiveness for an age band.

Participant ages · source passage

Abstract Background Early identification of neurodevelopmental conditions during the pre-school years (ages 3–6) is critical for timely intervention, yet traditional questionnaire-based screeners suffer from low positive predictive value, late detection, and inequitable access.

Study methods · source passage

This scoping review follows Joanna Briggs Institute methodology and the PRISMA-ScR checklist (Tricco et al., 2018); the protocol was registered on the Open Science Framework prior to charting. Inclusion criteria were (i) publication between 1 January 2020 and 15 June 2026; (ii) participants aged 72 months or younger at study entry; (iii) an explicit AI/ML/deep-learning screening or risk-prediction component; (iv) English-language peer-reviewed primary studies; and (v) reporting of at least one diagnostic-accuracy metric (sensitivity, specificity, AUC, PPV, NPV) or a prospective real-world deployment outcome. Reviews, regulatory documents, and foundational epidemiological/guideline references were retained separately for context and comparison but were not charted as primary tool studies. Exclusion criteria covered conference abstracts without a full paper, editorials/letters, method/benchmark papers on pre-labeled datasets without a target-population validation, adult or school-aged-only cohorts, and non-peer-reviewed pre-prints. Seven databases were searched (PubMed/MEDLINE, Web of Science Core Collection, Scopus, IEEE Xplore, ERIC, Cochrane CENTRAL, ClinicalTrials.gov), supplemented by citation searching of landmark papers and a hand search of the FDA 510(k)/De Novo database (product code QPF). Full search strings appear in the Supplementary material.

Limitations and uncertainty · source passage

Four limitations qualify this review. First, and most importantly for interpretation, the search was restricted to English-language peer-reviewed primary studies. This restriction carries a language bias whose direction is predictable rather than random: tools developed and validated in non-English-speaking settings—for example in China, Japan, Latin America, and Türkiye—are disproportionately likely to be reported in national-language journals and in databases we did not search, and are therefore under-represented in our map. Because one of our central findings is the geographic concentration of the evidence base, this bias may partly manufacture the very pattern we report, and the true modality and geographic breadth of the field is likely wider than Figure 4 suggests. Two considerations partially mitigate, but do not remove, this concern: several included tools originate from non-English-speaking countries (South Korea, Switzerland, France, Israel, China, Vietnam) yet were published in English, and regulatory milestones are documented in English-language registries. Future reviews should nonetheless search national-language sources (e.g., CNKI, SciELO, J-STAGE, ULAKBİM TR Dizin) with native-language strings, and we would expect such a search to soften—though probably not overturn—our conclusion about geographic concentration.

Design the system before the pilot

Use this study to inform a learning mechanism or implementation question. It is not direct evidence that autonomous agents improve learning. Keep the proposed agent behavior separate from what the source actually tested.

  • Co-design accommodations with learners and specialists. Check access and error burden across groups without inferring subgroup effectiveness from small or unreported samples.

A 90-day district pilot

  1. Days 1–15 · Appraise. Have the curriculum lead and a teacher read the methods and limitations. Record the study design, sample, comparator, outcomes and evidence gaps. Define a local question that the study can help inform.
  2. Days 16–30 · Design. Use a brief adult-mediated story, turn-taking or language activity. An educator controls every interaction; the agent never becomes an unsupervised companion. Preserve hands-on play and human conversation. Name an accountable owner and set the baseline, success threshold and budget before implementation.
  3. Days 31–60 · Test. Start with a small voluntary cohort, preserve an accessible non-AI route and require teacher review. Use synthetic or de-identified examples; approve data handling and tool terms through district processes.
  4. Days 61–90 · Decide. Measure independent naming, explanation or turn-taking in a human-led activity after the tool is removed. Record frustration, overstimulation and the amount of adult help. Compare with baseline, document uncertainty and stop or revise if learning worsens, access is unequal or critical errors persist.

The board decision brief

Report the learning problem, evidence quality, participating population, total cost including review time, measured benefit, errors, access differences and remaining uncertainties. Make a dated recommendation to stop, revise or expand. A positive perception survey alone does not establish learning improvement.

Attribution and reuse

Adem Arslan; Fatih Gül. Artificial-intelligence–based developmental screening for pre-school children (ages 3–6): a scoping review of deployed tools, clinical validity, equity, and implementation in guidance-center settings. Frontiers in Psychology (2026-08-20). https://doi.org/10.3389/fpsyg.2026.1913332

Companion prepared 2026-10-30. Original authorship and DOI are preserved. The abstract is an attributed source extract; the implementation plan is new AI-assisted editorial material, offered under the same CC BY 4.0 license. This guide has not undergone independent peer review.