Research Information System
European Journal of Oncology Nursing, vol.77, 2025 (SCI-Expanded, SSCI, Scopus)
Purpose: This prospective comparative study aimed to evaluate the performance of ChatGPT-4o in answering questions about endometrial cancer, with a focus on accuracy, empathy, and completeness in comparison to gynecologic oncologist. Additionally, it sought to explore potential differences in AI-generated responses between primary and secondary care settings, providing insights into its role at different levels of healthcare. Method: A total of 100 questions relevant to endometrial cancer were selected and validated by specialists. Fifty questions pertained to primary care (e.g., risk factors and prevention) and fifty to secondary care (e.g., diagnosis and treatment). Both ChatGPT-4o and a gynecologic oncologist answered these questions. Responses were evaluated by two independent oncologists based on accuracy, empathy, completeness. Also readability scores and word counts are calculated. Results: ChatGPT achieved significantly higher scores in accuracy (3.86 vs. 3.36; p < 0.001), empathy (3.47 vs. 1.66; p < 0.001), and completeness (3.00 vs. 1.97; p < 0.001) than the oncologist. Its answers, however, were much longer (403.51 vs. 25.06 words). Readability scores indicated that both ChatGPT and physician responses were similarly challenging to comprehend, requiring a high level of literacy. Conclusion: ChatGPT-4o demonstrated superior accuracy, empathy, and completeness in addressing questions about endometrial cancer compared to gynecologic oncologist. However, its lengthy responses could be overwhelming for patients. These findings suggest AI tools like ChatGPT-4o could support oncology nurses in delivering accurate patient education and empathetic communication, enhancing supportive care in endometrial cancer management. Future efforts should balance detail with clarity, enhancing AI's practical applicability in healthcare.