Toward Transparent AI-Enabled Patient Selection in Cosmetic Surgery by Integrating Reasoning and Medical LLMs

Area of Science:

• Artificial Intelligence in Medicine
• Clinical Decision Support Systems
• Healthcare Informatics

Background:

• Existing AI preoperative screening tools lack transparency and rely on fixed inputs.
• Current methods often use opaque feature weighting, limiting trust and interpretability.
• There is a need for AI solutions that offer clear reasoning and validation against clinical guidelines.

Discussion:

• A novel hybrid AI framework integrates reasoning large language models (LLMs) with specialized medical LLMs for enhanced preoperative screening.
• Patient data, both structured and free-text, is securely processed via a mobile app and a retrieval-augmented pipeline.
• The framework ensures transparency through chain-of-thought reasoning and validates risk factors against clinical guidelines using medical LLMs.

Key Insights:

• The hybrid AI model provides a composite suitability score with a detailed audit trail, enhancing clinical trust.
• It addresses critical challenges including model recency, hallucination control, data privacy, and algorithmic fairness.
• This approach offers a more interpretable and reliable alternative to current AI screening methods.

Outlook:

• Recommends a medical-device regulatory pathway for AI tools, emphasizing independent validation and ongoing bias monitoring.
• Proposes co-design with multidisciplinary stakeholders to ensure clinical relevance and ethical implementation.
• Future work should focus on real-world clinical integration and long-term performance evaluation.