Open Bite Classification Using Machine Learning: A Cephalometric Analysis.

Area of Science:

• Orthodontics
• Machine Learning
• Craniofacial Biology

Background:

• Anterior open bite (AOB) presents diagnostic challenges due to complex vertical craniofacial growth and skeletal patterns.
• Conventional cephalometric analysis alone is insufficient for objective AOB diagnosis.
• Machine learning (ML) offers advanced tools for orthodontic phenotypic characterization and diagnostic accuracy.

Purpose of the Study:

• To develop and evaluate an ML-based decision tree classifier for distinguishing AOB from healthy controls using cephalometric parameters.
• To explore latent craniofacial phenotypes within an Arab population using unsupervised clustering.
• To assess the potential of ML as a decision-support tool in orthodontic diagnosis and treatment planning.

Main Methods:

• Retrospective analysis of lateral cephalometric records from 1056 orthodontic patients (621 AOB, 435 controls) from the Arab population in Israel.
• Evaluation of five vertical skeletal cephalometric parameters: ML-NSL, NL-NSL, PFH/AFH, gonial angle, and facial axis.
• Development of a decision tree classifier and application of agglomerative hierarchical clustering for phenotype exploration.

Main Results:

• The decision tree classifier achieved 96.2% test accuracy, with high precision, recall, and F1-score (~0.97).
• The mandibular plane angle (ML-NSL) was the most influential feature for classification.
• Clustering identified ten distinct craniofacial clusters, including pure and mixed phenotypes, revealing varying vertical skeletal imbalances.

Conclusions:

• ML applied to cephalometric data provides accurate classification and phenotypic stratification of anterior open bite malocclusion.
• Clustering analysis uncovers clinically relevant subgroups, reflecting diverse vertical skeletal patterns.
• Interpretable ML models show promise as decision-support tools for personalized orthodontic diagnosis and treatment planning.