Evaluating artificial intelligence in decision-making for surgical treatment of benign breast conditions
Laura De Pellegrin1, Andrea Weinzierl2, Elisabeth A Kappos3
1Department of Plastic, Reconstructive and Aesthetic Surgery, Ospedale Regionale di Lugano, Ente Ospedaliero Cantonale (EOC), Lugano, Switzerland; Department of Plastic and Hand Surgery, Inselspital University Hospital Bern, University of Bern, Bern, Switzerland.
Introduction:
Determining the ideal surgical strategy for breast pathologies can sometimes be challenging, in particular if personalized solutions are required. Moreover, acquiring the necessary expertise takes years. Artificial intelligence (AI) may help streamline decision-making and improve treatment approaches. This study assessed the potential role of AI in therapeutic planning for common breast pathologies.
Methods:
Five clinical cases representing common breast pathologies were presented to ChatGPT-4o, an advanced AI model. Its responses were evaluated by 16 board-certified plastic surgeons and 9 residents for accuracy, relevance, and completeness using a Likert scale. Readability scores were also assessed by considering word length, syllable count, and sentence complexity.
Results:
AI-generated responses were found to be medically accurate, well-structured, and comprehensible. However, they lacked depth in surgical planning and risk assessment. The readability scores were Flesch Reading Ease (27.9), Flesch-Kincaid Grade Level (14.3), and Coleman-Liau Index (15.0), indicating a relatively high reading level.
Conclusion:
ChatGPT-4o can assist in analyzing breast pathologies; however, it does not account for individual patient factors and surgical nuances. Although ChatGPT-4o is not yet suitable for independent decision-making, further refinement, training, and expert validation could enhance the AI's role in future clinical applications.
