ESPE Yearbook of Paediatric Endocrinology

NPJ Digit Med. 2025;8(1):3. doi.org/10.1038/s41746-024-01382-4

Brief Summary: In this modeling study the cost-effectiveness of implementing autonomous artificial intelligence (AI) for diabetic retinal disease (DRD) screening in pediatric patients was evaluated, comparing it to traditional eye care provider (ECP) exams from a U.S. health system perspective. The analysis shows that AI screening becomes cost-saving when a site screens at least 241 patients annually, with larger health systems benefiting more due to economies of scale. AI-based screening consistently results in more patients being screened and adhering to follow-up care, with improved cost-effectiveness as system size increases.

A recent review and meta-analysis evaluated the diagnostic performance of deep learning (DL) algorithms applied to optical coherence tomography (OCT) and retinal images for detecting DRD (1). Analyzing 47 studies, the authors found that DL models achieved high accuracy and sensitivity, with pooled odds ratios indicating significant improvements over traditional methods. The findings support DL’s potential as a reliable, scalable tool for DRD screening to enhance early diagnosis and treatment, especially in resource-limited settings.

The study by Ahmed et al. demonstrates that AI screening can improve access, adherence, and equity in pediatric DRD care. Yet, the results are based on first-year implementation costs and may not reflect long-term savings. Also, some assumptions (e.g., patient behavior, cost estimates) may not generalize across all healthcare settings.

Thus, future studies should explore long-term cost-effectiveness, including downstream clinical outcomes. More granular analyses across diverse demographic, geographic, and insurance contexts are also needed. Additional research is required to assess AI effectiveness in routine care and in non-endocrine care settings.

Reference: 1. Deep learning-based optical coherence tomography and retinal images for detection of diabetic retinopathy: a systematic and meta analysis. Bi Z, Li J, Liu Q, Fang Z. Front Endocrinol (Lausanne). 2025;16:1485311. DOI 10.3389/fendo.2025.1485311