The effect of reference sample composition and size on dental age interval estimates.

Area of Science:

• Forensic Anthropology
• Pediatric Dentistry
• Biometrics

Background:

• Juvenile dental age estimation validation often overlooks interval performance across diverse ancestry groups.
• Limited research exists on how reference sample composition affects age interval accuracy.

Purpose of the Study:

• To investigate the impact of reference sample size and composition (sex, ancestry) on dental age interval estimates in children.
• To evaluate the performance of age estimation models under varying demographic conditions.

Main Methods:

• Utilized dental scores from 3334 London children (2-23 years) of Bangladeshi and European ancestry.
• Assessed model stability using standard error and tested age estimation performance with Bayesian multivariate cumulative probit and 5-fold cross-validation.
• Varied reference sample size, sex, and ancestry stratification.

Main Results:

• Decreasing sample size significantly increased standard error and negatively impacted performance metrics.
• Estimating age with mismatched sex reference and target samples reduced success rates.
• Ancestry group mixing had a lesser effect than sex mixing, with combined ancestry samples yielding comparable or better results than smaller single-demographic references.

Conclusions:

• Reference sample size and sex are primary drivers of dental age estimation performance.
• Combining reference samples by ancestry enhances age estimation accuracy, supporting population specificity over broad demographic assumptions.