Machine Learning-Based Model for Predicting Short- and Long-Term Growth in Untreated Class III Malocclusion

Area of Science:

• Orthodontics and Dental Anthropology
• Biomedical Engineering
• Machine Learning in Healthcare

Background:

• Class III malocclusion presents complex growth patterns in Caucasian individuals.
• Predicting skeletal growth is crucial for effective orthodontic treatment planning.
• Existing methods for growth prediction have limitations in accuracy and scope.

Purpose of the Study:

• To develop and evaluate a Machine Learning (ML)-based model for predicting the craniofacial growth of Caucasian subjects with untreated Class III malocclusion.
• To assess the model's predictive accuracy in both short-term and long-term growth phases.
• To identify specific cephalometric landmarks with varying prediction accuracies.

Main Methods:

• Utilized a longitudinal sample of 144 Caucasian subjects with untreated Class III malocclusion.
• Employed a Graph Neural Network (GNN) model trained on 80% of the data.
• Analyzed cephalograms using 16 digitized cephalometric landmarks in an X-Y coordinate system.
• Validated predictions using a one-sample t-test and calculating Euclidean distances between predicted and observed values on a 20% test set.

Main Results:

• The ML model demonstrated statistically significant predictions for several cephalometric points (SX, PgY, BY, PNSY, NY, SY in short-term; MeX, GnX, PgX, BX, BY, PtY in long-term).
• In short-term predictions, high mean Euclidean distances (indicating lower accuracy) were observed for mandibular landmarks: Go (2.6 mm), Me (1.9 mm), Gn (1.9 mm), Pg (2.0 mm), and B (2.0 mm).
• Long-term predictions also showed high mean Euclidean distances for mandibular landmarks: Go (3.1 mm), Me (4.3 mm), Gn (4.1 mm), Pg (4.5 mm), and B Point (4.1 mm).

Conclusions:

• The developed ML model shows promising accuracy for predicting craniofacial growth in Class III malocclusion across most landmarks.
• Prediction accuracy is notably lower for key mandibular landmarks (Go, Me, Gn, Pg, B Point) in both short-term and long-term analyses.
• Further refinement of ML models may be necessary to improve the prediction of mandibular growth in this patient cohort.