Enhanced Bone Age Evaluation Model for Chinese Children Aged 3-18 Years
Overview
A simplified bone age assessment (BAA) model using three key bones (radius, ulna, and first metacarpal) was developed from 5551 Chinese children aged 3 to 18 years. This model demonstrated high accuracy (R² > 0.94) and low error (<0.5 years), offering a rapid and reliable alternative to traditional complex methods, especially when incorporating pubertal staging.
Background
Bone age assessment is vital for evaluating physiological maturity and diagnosing growth disorders in children. Traditional methods like the Tanner-Whitehouse 3-China (TW3-C) system, while accurate, are time-consuming and require expert interpretation. The TW3-C-RUS method, focusing on 13 radius-ulna-short bones, is commonly used in Chinese pediatric populations but involves complex scoring. Simplifying BAA by identifying key bones and directly modeling bone grades to bone age could reduce information loss and improve clinical efficiency.
Data Highlights
Parameter
Value
Sample Size
5551 children aged 3-18 years
Model Type
Linear regression with bone grades as predictors
Key Bones in Simplified Model
Radius, Ulna, Metacarpal I
Model R²
> 0.94
Root Mean Square Error
< 0.5 years
Bone Age Error Margin for Accuracy
≤ 0.5 years
Key Findings
The simplified 3-bone model (radius, ulna, metacarpal I) achieved high accuracy with R² > 0.94 and RMSE < 0.5 years.
When bone grades were consistent, single or few bones (e.g., metacarpals and phalanges at grade 6) could approximate bone age effectively.
In cases of inconsistent bone grades, regression models using multiple bones improved assessment accuracy.
Incorporating pubertal staging into the models further enhanced bone age prediction accuracy.
The TW3-C-RUS method remains the gold standard but is time-consuming and requires expertise, motivating the development of simplified models.
Large-scale cross-sectional data from Beijing children provided a robust basis for model development and validation.
Clinical Implications
The simplified 3-bone regression model offers pediatricians a rapid, reliable tool for bone age assessment without extensive scoring or software. Incorporating pubertal stage information can further refine accuracy, aiding clinical decisions in growth and puberty-related disorders. This approach may reduce variability due to physician experience and streamline workflow in busy clinical settings.
Conclusion
This study presents a clinically practical, simplified bone age assessment model for Chinese children that maintains high accuracy while reducing complexity. Further validation across diverse populations is needed to confirm its generalizability and robustness.
References
Development of an Enhanced Model for Bone Age Evaluation in Chinese Children Aged 3 to 18 Years
