Related Experiment Video
Updated: Jul 12, 2026
08:03
Midface Hypoplasia and Cranial Base Morphology in Syndromic Craniosynostosis: A Comparative Analysis Study Using a Predictive Regression Model
Published on: November 4, 2025
Integrated framework for pediatric height assessment: X-ray-based height extreme cases classification and machine
Ya-Wen Chang1, Meng-Che Tsai2, Sun-Yuan Hsieh3,4,5,6,7
1Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, 701, Taiwan ROC.
Scientific Reports
|July 9, 2026
Summary
This study introduces a new AI framework to predict children's future height abnormalities using hand X-rays and clinical data. The system accurately identifies extreme height cases and forecasts growth trajectories for better pediatric care.
Area of Science:
- Pediatric endocrinology and growth assessment.
- Artificial intelligence in medical diagnostics.
- Biometric data analysis for human growth.
Background:
- Accurate pediatric height assessment is vital for monitoring growth and timely interventions.
- Traditional methods often focus on current bone age, limiting prediction of future height deviations.
- There is a need for advanced tools to predict potential extreme height outcomes in children.
Purpose of the Study:
- To develop and validate an integrated AI framework for predicting pediatric height abnormalities.
- To identify children at risk for extreme short or tall stature.
- To forecast height trajectories across various time scales for improved clinical decision-making.
Main Methods:
- A two-component framework utilizing hand X-ray images and clinical data.
- Inception-ResNet-V2 model for classifying potential extreme height cases, with DSEV for data imbalance.
- XGBoost model for multivariate height prediction using anthropometric and medical features.
- Evaluation of prediction accuracy at 6 months, 1 year, 2 years, and near-final adult height.
Main Results:
- Classification models demonstrated high accuracy and reliability in detecting unexpectedly short and tall children.
- Multivariate prediction models achieved very low mean absolute error for short-term and long-term height predictions.
- The framework showed consistent and stable performance across different prediction horizons.
Conclusions:
- The integrated AI framework offers precise and comprehensive tools for pediatric growth evaluation.
- This approach enhances the ability to predict and manage future height abnormalities.
- It supports clinicians in making informed decisions for growth monitoring and intervention planning.
