[Advances in prediction models for temporomandibular disorders]

Area of Science:

• Oral and Maxillofacial Surgery
• Computational Biology
• Data Science

Background:

• Temporomandibular disorders (TMD) are prevalent, significantly impacting patient quality of life.
• Accurate prediction and timely treatment of TMD are essential in oral and maxillofacial surgery.
• The evolution of TMD prediction models reflects advancements in computational methodologies.

Purpose of the Study:

• To comprehensively review and evaluate computational approaches for TMD prediction.
• To critically analyze the strengths and limitations of statistical, machine learning, and deep learning models in TMD prediction.
• To discuss future research directions for improving TMD prediction accuracy and efficacy.

Main Methods:

• Review of traditional statistical methods for identifying risk factors in TMD.
• Analysis of machine learning techniques for pattern recognition in large-scale TMD datasets.
• Evaluation of deep learning approaches for temporal pattern extraction and nonlinear relationship modeling in TMD time-series data.

Main Results:

• Traditional methods offer interpretability but require prior assumptions.
• Machine learning excels with high-dimensional data but depends on data quality and generalization.
• Deep learning captures complex temporal dynamics but needs large datasets and faces interpretability challenges.

Conclusions:

• Each computational approach presents distinct advantages and limitations for TMD prediction.
• Understanding these trade-offs is key to selecting appropriate models for clinical application.
• Further research is needed to enhance the accuracy, generalizability, and interpretability of TMD prediction models.