The abstract is read at the publisher; the summary is JClub's.
Fluorescence lifetime imaging ophthalmoscopy (FLIO)–derived center-of-mass mean lifetime maps classify normal, prediabetic, and diabetic status with a CNN AUROC of 0.60 ± 0.04.
- Why it matters: Retinal metabolic changes linked to type 2 diabetes may be detectable noninvasively, but it is unclear whether FLIO signals can support clinically useful deep-learning classification beyond age.
- What they did: In a cross-sectional AI-READI dataset (version 3) of 1,783 participants with HbA1c, the study generated corrected tauCMM maps from time-resolved FLIO scans and trained CNN, ResNet-18, and XGBoost for three-class and binary tasks.
- The result: tauCMM was longer in prediabetic and diabetic groups than in normoglycemic participants (p < 0.001), with the SSC difference remaining after age adjustment (p = 0.017); FLIO models exceeded age-only across tasks, but adding age did not improve performance.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.