Clustering and Machine Learning Models of Skeletal Class I and II Parameters of Arab Orthodontic Patients.

Background

Orthodontic problems can affect vital quality of life functions, such as swallowing, speech sound production, and the aesthetic effect. Therefore, it is important to diagnose and treat these patients precisely. The main aim of this study is to introduce new classification methods for skeletal class I occlusion (SCIO) and skeletal class II malocclusion (SCIIMO) among Arab patients in Israel. We conducted hierarchical clustering to detect critical trends within malocclusion classes and applied machine learning (ML) models to predict classification outcomes.

Methods

This study is based on assessing the lateral cephalometric parameters from the Center for Dentistry Research and Aesthetics based in Jatt, Israel. The study involved the encoded records of 394 Arab patients with diagnoses of SCIO/SCIIMO, according to the individualized ANB of Panagiotidis and Witt. After clustering analysis, an ML model was established by evaluating the performance of different models.

Results

The clustering analysis identified three distinct clusters for each skeletal class (SCIO and SCIIMO). Among SCIO clusters, the results showed that in the second cluster, retrognathism of the mandible was less severe, as represented by a lower ANB angle. In addition, the third cluster had a lower NL-ML angle, gonial angle, SN-Ba angle, and lower ML-NSL angle compared to clusters 1 and 2. Among SCIIMO clusters, the results also showed that the second cluster has less severe retrognathism of the mandible, which is represented by a lower ANB angle and Calculated_ANB and a higher SNB angle (p < 0.05). The general ML model that included all measurements for patient classification showed a classification accuracy of 0.87 in the Random Forest and the Classification and Regression Tree models. Using ANB angle and Wits appraisal only in the ML, an accuracy of 0.78 (sensitivity = 0.80, specificity = 0.76) was achieved to classify patients as SCIO or SCIIMO.

Conclusions

The clustering analysis revealed distinguished patterns that can be present within SCIO and SCIIMO patients, which can affect the diagnosis and treatment plan. In addition, the ML model can accurately diagnose SCIO/SCIIMO patients, which should improve precise diagnostics.