Large Language Models Diagnose Facial Deformity.

Area of Science:

• Medical Diagnostics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Traditional methods for diagnosing jaw deformities have limitations in data interpretation and accessibility.
• Advanced computational tools are needed to streamline complex cephalometric analysis.

Purpose of the Study:

• To investigate the application of Large Language Models (LLMs) for diagnosing jaw deformities.
• To enhance data interpretation and accessibility in clinical diagnostic processes.

Main Methods:

• Cephalometric measurements from patients with jaw deformities were converted into text for LLM analysis.
• Multiple LLMs (LLAMA-2, GPT, Gemini-Pro) were evaluated against threshold-based and machine learning models.
• Performance was assessed using balanced accuracy and F1-score.

Main Results:

• Larger LLMs demonstrated efficient adaptation to diagnostic tasks with minimal training data.
• LLMs reduced classification ambiguity, showcasing strong in-context learning.
• Text conversion of measurements improved interpretability and provided actionable clinical insights.

Conclusions:

• Integrating LLMs into jaw deformity diagnosis significantly improves accessibility and reduces reliance on specialized training.
• LLMs serve as valuable auxiliary tools, simplifying decision-making for clinicians.
• Future work with larger, medically specific datasets will further enhance LLM precision and utility in diagnostics.