We also have X and podcasts
Smart Screening: AI-Enhanced Risk Model Reduces ROP Examinations While Maintaining Safety
A new validated risk model combining gestational age, postmenstrual age, and vascular severity assessment can predict treatment-requiring retinopathy of prematurity (TR-ROP) within 2 weeks, potentially reducing screening examinations by up to 39% without missing critical cases.
Background
Retinopathy of prematurity screening requires frequent, serial eye examinations in premature infants, imposing cumulative stress and medical burden. Current screening protocols rely primarily on chronological markers without incorporating quantitative assessment of vascular development, potentially leading to unnecessary examinations.
Key Findings
- Adding vascular severity assessment to gestational and postmenstrual age improved discrimination significantly (AUROC improvement 0.13 in training cohort, 0.08 in external validation)
- The logistic regression model achieved 100% sensitivity for detecting 2-week TR-ROP risk with 63–73% specificity across cohorts
- Risk-based scheduling simulations reduced examinations by 28–39% without missing any treatment-requiring cases
- Clinician-assigned severity scores performed comparably to AI-derived scores (AUROC 0.87 with 100% sensitivity), supporting practical implementation
Why It Matters
This model demonstrates that quantitative vascular assessment, beyond chronological milestones, substantially predicts disease progression. Risk-based scheduling could reduce unnecessary examinations, decreasing medical burden on vulnerable infants while reducing healthcare costs—without compromising safety.
Limitations
The study was retrospective. Prospective implementation within defined clinical workflows is required before routine adoption. External validation across additional populations is needed.
Original paper: Precision Risk Model Using Quantitative Assessment of Vascular Severity in Telemedicine-Based Screening. — JAMA ophthalmology. 10.1001/jamaophthalmol.2026.0510
