Artificial Intelligence-Based Approach for Determining the Risk of Temporomandibular Disorders.

Area of Science:

• Biomedical Informatics
• Artificial Intelligence in Healthcare
• Dental Research

Background:

• Temporomandibular Disorders (TMD) present diagnostic challenges.
• Early detection and risk assessment are crucial for effective management of TMD.

Purpose of the Study:

• To predict TMD using machine learning (ML) approaches.
• To leverage clinical and sociodemographic variables for early TMD detection and risk stratification.

Main Methods:

• A retrospective-prospective study involving 340 patients with and without TMD.
• Utilized five ML models: DANets, KAN, KNN, MLP, and SVM.
• Evaluated performance using accuracy, precision, recall, F1-score, and SHAP for interpretability.

Main Results:

• The DANets model demonstrated superior performance in both result (accuracy=0.8529) and diagnostic classification (accuracy=0.8382, F1=0.8212).
• SHAP analysis identified age, gender, joint pain, clicking, and limited mouth opening as key predictive factors.

Conclusions:

• AI, particularly deep learning, can significantly aid clinicians in TMD diagnosis and risk assessment.
• These models can reduce diagnostic uncertainty, facilitate early intervention, and potentially prevent TMD progression.