Nine Years of Pediatric Iris Recognition: Evidence for Biometric Permanence from Childhood through Adolescence

Large-scale identity systems deploy iris recognition for minors, including national identification and border control applications, yet re-enrollment policies lack longitudinal validation, creating uncertainty regarding authentication reliability as enrollees mature. To address this gap, we conducted a longitudinal study of 274 children followed for up to 9 years (ages 4-19), comprising 18,318 iris images. Using linear mixedeffects models with random slopes, we evaluated both commercial and open-source iris matchers while disentangling enrollment age, developmental maturation, and elapsed time. False nonmatch rates remained stable at 0.018% and 0.024% across the observation period for the evaluated systems. When individual heterogeneity was explicitly modeled, no statistically significant population-level temporal effects were observed (Wald tests, all p > 0.05), indicating that apparent degradation in randomintercept-only models (which do not account for individual heterogeneity) arises from heterogeneous individual trajectories rather than systematic biometric aging. Between-subject differences accounted for 72-85% of total variance (intraclass correlation coefficient = 0.72-0.85), while image quality and pupil dilation consistency produced substantially larger effects (1.5-2.0 standard deviations) than temporal factors. All false non-matches were concentrated in a small subset of subjects with persistent acquisition challenges. These findings demonstrate that pediatric iris recognition maintains authentication accuracy comparable to adult benchmarks through adolescence, supporting qualitytriggered re-enrollment strategies over fixed temporal intervals in secure identity systems.