Ophthalmology

Smartphone AI detects ocular surface malignancies in community self-screening

In a nonrandomized Chinese trial of 614 at-home screeners, a smartphone AI app achieved smartphone AUC 0.905 versus slitlamp AUC 0.945 for malignant versus benign lesions. Population-level performance reported AUC 0.977, sensitivity 89.3%, and specificity 95.9% during outreach-driven screening. The app identified 20 pathologically confirmed malignancies, with 19 newly diagnosed and no enucleations required. This tool can augment early referral pathways, but clinic confirmation and pathology remain required for management decisions. 1

High-intensity evening light linked to higher risk of AMD, cataract, and POAG

In 82,826 UK Biobank participants followed median 7.85 years, evening light >1000 lx associated with higher hazards of AMD (HR 1.31), cataract (HR 1.18), and POAG (HR 1.47). A dose-response was reported: each hour >2250 lx increased overall ARED risk (HR 1.10) and POAG (HR 1.18). Exposure was measured by wrist-worn sensors over 7 days, and outcomes used hospital diagnostic codes. Observational design supports modifiable-risk hypothesis but cannot prove causation, and residual confounding is possible. 2

Low-dose atropine (0.01%) is likely cost-effective for UK pediatric myopia

A UK lifetime model found SYD-101 (0.01%) added to standard care increased QALYs by 0.139 at incremental cost £2239, yielding ICER £16,164/QALY. Probabilistic sensitivity analysis showed moderate probability of cost-effectiveness (≈58.2%) at conventional thresholds. Benefits were driven by reduced progression to high myopia and fewer complications in modeled lifetime horizons. Applicability is to children meeting trial-like progression criteria in the UK; real-world uptake and adherence may alter cost-effectiveness. 3