ChatGPT vs Expert-Guided Care Pathways for Postesophagectomy Symptom Management

Area of Science:

• Oncology
• Medical Informatics
• Artificial Intelligence

Background:

• Esophageal cancer surgery (esophagectomy) presents complex symptom management challenges.
• Expert-guided care pathways are crucial for post-operative patient care.
• Generative AI tools are emerging as potential aids in clinical decision-making.

Purpose of the Study:

• To compare generative AI (ChatGPT) care pathways with expert-guided (AskMayoExpert) pathways for managing post-esophagectomy symptoms.
• To evaluate the comprehensiveness and clinical utility of AI-generated symptom management strategies.

Main Methods:

• Developed 9 expert-guided care pathways (AskMayoExpert) for post-esophagectomy domains (e.g., dysphagia, pain, mental health).
• Utilized ChatGPT (versions 3.5 and 4.0) to generate responses to similar clinical questions.
• Compared AI-generated pathways with expert pathways using a 5-question survey tool.

Main Results:

• Both AI and expert pathways defined symptoms and management options for all 9 domains.
• ChatGPT 3.5, with simple prompts, lacked stepwise approaches, testing recommendations, and triage.
• ChatGPT 4.0, with sophisticated prompts, generated pathways comparable to expert guidance.

Conclusions:

• Generative AI can serve as a supplementary tool for physicians managing complex post-esophagectomy symptoms.
• Current AI technology is insufficient for sole clinical management of severe or complex patient cases.
• Continued advancement in AI is expected to enhance its role in clinical practice.