Comparison of Artificial Intelligence Responses About Penile Prosthesis implantation in Terms of Quality and Readability: A Methodological and Validity Study on ChatGPT, Gemini, and DeepSeek

Objective: The use of artificial intelligence models to inform patients about penile prosthesis implantation has not been sufficiently studied in the literature. This study evaluates the quality and readability of responses given by ChatGPT, DeepSeek, and Gemini to the most common questions asked by patients before penile prosthesis implantation. Material and Methods: We selected 15 common patient questions related to preoperative management, complications, and postoperative care regarding penile prosthesis implantation. These questions were submitted in English on February 10, 2025, to ChatGPT, DeepSeek, and Gemini. Responses were analyzed for information quality (using the DISCERN Tool) and readability (using the Gunning Fog Index). Results: ChatGPT had the highest DISCERN score (4.2±0.3), while DeepSeek (3.8±0.4) and Gemini (3.5±0.5) had lower scores (p<0.05 for all comparisons). In terms of readability, ChatGPT generated the most comprehensible responses (Gunning Fog Index: 7.1±0.6), whereas DeepSeek (8.3±0.7) and Gemini (9.0±0.8) produced more complex texts (p<0.05 for all comparisons). Conclusion: ChatGPT outperformed the other models in terms of information quality and readability, demonstrating its potential as a complementary tool for patient education and informed consent. However, the observed differences in information accuracy and readability among the models highlight the need for further refinement before full integration into clinical use.