Peer-reviewed article
Clinical Performance and Implementation of AI-Enabled Paediatric Ophthalmic Screening, Triage, Diagnosis, and Surveillance in Primary, Community, and Referral-Linked Pathways: A Systematic Review.
Somerville J, Mustafa MH, Seweid MM, El Arab RA ·
Summary
Written by AI, not yet AICS's view · claude-haiku-4-5-20251001This systematic review examines artificial intelligence tools designed to screen, diagnose, and monitor eye conditions in children aged 0-18 years. The review included 12 studies and one dataset across conditions like retinopathy of prematurity, retinoblastoma, myopia, and amblyopia. AI systems showed promising performance in detecting certain eye conditions and worked feasibly in community and school settings using smartphone imaging. However, the evidence remains limited by small studies, lack of testing across different populations and settings, and insufficient fairness assessments, so independent use of AI in child eye care is not yet ready.
Key points
- AI systems demonstrated promising diagnostic performance for detecting treatment-requiring retinopathy of prematurity, discriminating retinoblastoma activity, and predicting myopia from fundus images
- Several studies supported feasibility in real-world pathways including neonatal units, schools, and communities using smartphone-based imaging and task-shifted operators
- Critical limitations included single-centre studies with enriched samples, small numbers of clinically important cases, and heterogeneous study designs
- External validation, temporal validation, calibration, and fairness assessments across different patient subgroups were largely absent from the included studies
- Routine autonomous implementation of AI tools in paediatric eye care remains premature; human oversight and image-quality control are necessary safeguards
Limitations
The review does not evaluate the long-term health outcomes for children who received AI-assisted diagnosis or screening. It cannot determine which specific AI approaches are most cost-effective or implementable in resource-limited settings where paediatric eye care access is poorest.
Abstract
Background/Objectives: Artificial intelligence (AI) is increasingly being evaluated for ophthalmic diagnosis, screening, and triage, yet its role in paediatric eye care remains less established than in adult ophthalmology. This systematic review aimed to synthesise evidence on AI-enabled tools for paediatric ophthalmic diagnosis, screening, triage, surveillance, and referral, with an emphasis on diagnostic performance, safety, workflow integration, equity, and implementation readiness in primary, community, and primary care-relevant settings. Methods: A PRISMA-guided systematic review was conducted using MEDLINE, Embase, Web of Science, Scopus, and IEEE Xplore from inception to 30 March 2026. Eligible studies evaluated AI or machine-learning tools for children and adolescents aged 0-18 years in relation to paediatric eye conditions. Study selection and data extraction were undertaken independently by reviewers, with disagreements resolved by consensus or third-reviewer adjudication. Methodological and reporting quality was evaluated using an author-adapted six-domain rubric informed by APPRAISE-AI. Diagnostic-accuracy studies were assessed using an author-adapted QUADAS-2 framework incorporating QUADAS-AI-informed AI-specific considerations, the prediction-model study was assessed using PROBAST+AI, and the non-randomised treatment-effect study was assessed using ROBINS-I. The public dataset descriptor was evaluated separately using an author-developed dataset-quality, representativeness, and applicability framework. Because of clinical and methodological heterogeneity, findings were synthesised thematically. Results: Twelve empirical studies and one public dataset descriptor were included, covering retinopathy of prematurity, retinoblastoma, amblyopia risk, myopia, congenital cataract, and visual-acuity assessment. AI systems frequently demonstrated promising diagnostic or screening performance, including sensitivity-first detection of treatment-requiring retinopathy of prematurity, high discrimination for retinoblastoma activity, and strong myopia prediction using fundus images. Several studies supported feasibility in neonatal, school, and community workflows using smartphone-based imaging, task-shifted operators, tele-referral, and human-in-the-loop review. However, external and temporal validation, calibration, patient-level reporting, subgroup and fairness assessment, and economic evaluation were limited. Conclusions: AI-enabled tools show promise for supporting selected paediatric ophthalmic screening, triage, and surveillance pathways, particularly when combined with image-quality control, explicit escalation, and human oversight. However, confidence in the reported performance is limited by single-centre studies and enriched samples, small numbers of clinically important cases, heterogeneous analytical units, potentially optimistic aggregation procedures, limited external or temporal validation, incomplete calibration, and absent fairness analyses. Routine autonomous implementation remains premature.