An Interpretable Fuzzy-AI Clinical Decision Support System for Selecting Orthognathic Surgery in Skeletal Class III...

Area of Science:

• Orthodontics and Dental AI
• Skeletal Malocclusion Analysis
• Clinical Decision Support Systems

Background:

• Skeletal class III malocclusion presents complex treatment choices between orthognathic surgery and orthodontic camouflage.
• Accurate diagnosis and treatment planning are crucial for optimal patient outcomes.
• Existing decision-making processes can lack standardization and transparency.

Purpose of the Study:

• To develop an interpretable fuzzy-artificial intelligence (AI) framework for treatment decision-making in skeletal class III malocclusion.
• To integrate a large language model (LLM) for generating clinician-oriented explanations.
• To enhance the transparency and reproducibility of treatment planning.

Main Methods:

• A fuzzy inference model calculated a membership grade (MG) for skeletal class III severity using cephalometric variables.
• A decision-tree model was built using MG and other cephalometric data to predict treatment choice.
• An LLM translated the AI model's output into understandable explanations, creating an integrated system.

Main Results:

• The AI model achieved high discriminative performance with an area under the curve (AUC) of 0.97.
• Key predictors for treatment selection included overjet, MG, lower facial height, and the H angle.
• The LLM effectively generated concise, clinician-friendly diagnostic summaries.

Conclusions:

• The LLM-integrated fuzzy-AI framework successfully formalizes expert clinical reasoning for skeletal class III malocclusion.
• This approach enhances transparency, standardization, and reproducibility in treatment planning.
• The developed system is available as open-source software for broader clinical application.