The increasing use of artificial intelligence (AI) chatbots in healthcare is an important convenience brought about by the digital age (1). Especially individuals with limited access to health services or people who are hesitant to apply to health professionals use these AI programs widely. Pelvic floor muscle exercises are one of the non-pharmacological methods recommended for the management of pelvic floor dysfunctions (2). Applying the exercises with the wrong technique or inappropriate information may reduce expected clinical benefit and may also lead to an increase in symptoms. Due to the intimate nature of pelvic floor complaints and the limitations in accessing health care, individuals often turn to online information sources. So, it is important to evaluate the accuracy, consistency, and clinical appropriateness of the information presented on digital platforms. This study was conducted in order to compare the responses of two different AI chatbots (OpenAI ChatGPT 5.2 Thinking and Google Gemini 3 Pro) related to pelvic floor muscle exercises to investigate the reliability, consistency, and quality of information.

Firstly, on the Google Trends platform, searches conducted worldwide with reference to the term “pelvic floor muscle exercises” from 2004 to the present have been researched. Based on the keywords included in the highest volume of queries associated with this term, repetitive phrases were eliminated, and as a result, the 10 most frequently asked questions were compiled (Table 1). Each question compiled was directed to OpenAI ChatGPT 5.2 Thinking and Google Gemini 3 Pro chatbots using standardized prompts to ensure comparability of answers. To minimize session-induced bias and context transfer, each question was queried in a separate chat session (a new chatbot); in addition, queries were conducted via anonymous accounts that had not been used before. The readability of the answers was evaluated by the Flesch Ease of Legibility Score (FRES) and the Flesch–Kincaid Grade Level (FKGL). Information quality was investigated with the Ensuring Quality Information for Patients (EQIP) tool, and the dimensions of openness, accuracy, and reliability were examined with the modified DISCERN (mDISCERN) tool. All evaluations were carried out by two independent physiotherapists experienced in the field of pelvic floor rehabilitation. In cases where there is a difference of opinion between the evaluators, the final decision is made by the evaluation of a third physiotherapist.

Comparisons between the groups were conducted using the Mann–Whitney U test. It was found that Google Gemini 3-Pro responses were statistically significantly higher in terms of EQIP and mDISCERN total scores compared to OpenAI ChatGPT 5.2 Thinking responses (p=0.023). On the other hand, there was no significant difference between the two artificial intelligence chatbots in FRES and FKGL scores (p˃0.05) (Table 2).

In this study, when the answers to the questions about pelvic floor muscle exercises were evaluated, it was determined that the answers of Google Gemini 3 Pro were higher in terms of information quality and reliability compared to OpenAI ChatGPT 5.2 Thinking. In contrast, it was found that the responses of the two AI chatbots were at a similar readability level in terms of readability criteria.

In conclusion, it was determined that the responses of Google Gemini 3 Pro and OpenAI ChatGPT 5.2-Thinking to pelvic floor muscle exercises differed in terms of information quality and reliability, and Gemini 3 Pro performed better in these dimensions. However, it is important to have approaches aimed at increasing e-health literacy in order for patients to access accurate and reliable information. In addition, it may be useful to strengthen the interaction between clinicians and AI developers in the process of deconstructing the health content of AI systems and making them more clinically appropriate.

King, M. R. (2023). The future of AI in medicine: a perspective from a Chatbot. Annals of Biomedical Engineering, 51(2), 291-295.
Bo, K., Frawley, H. C., Haylen, B. T., Abramov, Y., Almeida, F. G., Berghmans, B., ... & Wells, A. (2017). An International Urogynecological Association (IUGA)/International Continence Society (ICS) joint report on the terminology for the conservative and nonpharmacological management of female pelvic floor dysfunction. International urogynecology journal, 28(2), 191-213.

Continence 19S (2026) 102809
DOI: 10.1016/j.cont.2026.102809