Answers are grounded in this paper and cite their sources.
Background Predicting nasal growth remains a major challenge in adolescent facial treatment. Current clinical decision-making relies primarily on chronological age and indirect skeletal maturity indicators, which poorly reflect individual nasal growth patterns. Three-dimensional facial imaging offers new possibilities for morphometric analysis; however, most existing growth models lack sufficient nasal resolution. This study aimed to clinically validate the geometric accuracy of a three-dimensional nasal growth prediction algorithm based on dense landmark-based morphometry. Methods The predictive model was developed using a cross-sectional dataset of approximately 2,050 multi-ethnic subjects aged 6–19 years with equal sex distribution. A total of 84 facial landmarks were collected, including 54 nasal-specific landmarks. A supervised multivariable linear regression model was trained on the standardised morphometric parameters, with age (continuous) and biological sex as covariates; no dimensionality reduction was applied. External validation was performed on a cohort of 12 longitudinally followed adolescents (6 female, 6 male; baseline 12.7 ± 2.6 years, follow-up 15.4 ± 2.7 years), strictly independent of the development dataset. The predicted post-pubertal morphology was compared with the observed follow-up acquisition using three complementary analyses: surface-to-surface deviation, Bland-Altman concordance on the Goode ratio and the nasal tip projection, and comparison against a no-growth baseline scenario. Results The mean surface deviation between predicted and observed nasal morphology was 0.58 ± 0.20 mm (95% CI 0.46–0.70 mm; RMSE = 0.61 mm). 95.6 ± 5.0% of nasal surface points lay within the 1.5-mm clinical threshold and 99.5 ± 0.9% within 2.5 mm. Bland-Altman concordance analysis confirmed near-perfect proportional agreement on the Goode ratio (bias = −0.033; 95% limits of agreement [−0.145; +0.079]) and sub-threshold systematic bias on the nasal tip projection (bias = −1.01 mm, below the 1.5-mm clinical threshold; 95% limits of agreement [−4.10; +2.09] mm). The algorithm consistently outperformed the no-growth baseline scenario on both parameters, with the SD of the differences reduced by 12.3% and 8.7% respectively. Conclusion The proposed three-dimensional algorithm provides geometric and morphometric validation of nasal growth prediction on a longitudinal adolescent cohort, with sub-millimetric global accuracy and concordant prediction on clinically relevant nasal parameters. These findings support the clinical applicability of the algorithm in the assessment of adolescent nasal growth, therapeutic planning and patient counselling, while accumulation of larger prospective datasets will continue to refine the precision boundaries across diverse populations.
No takes yet. Share an insight, caveat, or question.
Pozzo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: