International journal of sexual health : official journal of the World Association for Sexual Health · 2026
ChatGPT as a Patient Education Tool in Female Sexual Dysfunction: A Clinician-Based Evaluation.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Akdemir Celal, Balcı Mücahit Furkan, Yıldırım Fatih, Kıncı Mehmet Ferdi, Pay Ramazan Erda
Аннотация
OBJECTIVE: To provide a methodologically robust evaluation of ChatGPT® responses to frequently asked patient questions on female sexual dysfunction (FSD) using operationalized rating criteria and cluster-aware statistical analyses.
METHODS: This study was conducted in Türkiye. Ten commonly encountered FSD questions were identified through a structured review of patient encounters. ChatGPT® (March 2025) generated standardized responses using a fixed prompt. A multinational panel of obstetrics and gynecology specialists (n = 22) evaluated each response across three operationally defined criteria-accuracy, comprehensiveness, and safety-using a 5-point anchored Likert scale. The intentionally incorrect item was retained in the full dataset and also assessed separately as an attention check. Primary comparisons across criteria were performed using repeated-measures ANOVA, with mixed-effects sensitivity models including random intercepts for raters and questions. Secondary descriptive threshold analyses were performed using alternative positivity cut-points.
RESULTS: A total of 660 ratings were generated across ten ChatGPT® responses. The overall mean rating was 4.08 ± 0.27. Accuracy received the highest mean score (4.22 ± 0.26), followed by safety (3.96 ± 0.28) and comprehensiveness (3.86 ± 0.30). Repeated-measures ANOVA showed significant differences across the three criteria (p = 0.014), and mixed-effects modeling demonstrated the same directional pattern. Treatment-focused questions received higher ratings (4.19 ± 0.24) than diagnostic questions (3.75 ± 0.31). Inter-rater reliability was moderate to good across domains. The intentionally incorrect response was correctly identified as inaccurate and/or unsafe by 20 of 22 raters (90.9%).
CONCLUSION: In this clinician-based expert review, ChatGPT® showed relatively strong performance in accuracy but lower performance in comprehensiveness and variable performance in safety when addressing FSD-related questions. These findings suggest that ChatGPT® may serve as a supplementary patient-education tool for basic, guideline-aligned topics, but its outputs should be regarded as informational starting points rather than stand-alone clinical recommendations.
METHODS: This study was conducted in Türkiye. Ten commonly encountered FSD questions were identified through a structured review of patient encounters. ChatGPT® (March 2025) generated standardized responses using a fixed prompt. A multinational panel of obstetrics and gynecology specialists (n = 22) evaluated each response across three operationally defined criteria-accuracy, comprehensiveness, and safety-using a 5-point anchored Likert scale. The intentionally incorrect item was retained in the full dataset and also assessed separately as an attention check. Primary comparisons across criteria were performed using repeated-measures ANOVA, with mixed-effects sensitivity models including random intercepts for raters and questions. Secondary descriptive threshold analyses were performed using alternative positivity cut-points.
RESULTS: A total of 660 ratings were generated across ten ChatGPT® responses. The overall mean rating was 4.08 ± 0.27. Accuracy received the highest mean score (4.22 ± 0.26), followed by safety (3.96 ± 0.28) and comprehensiveness (3.86 ± 0.30). Repeated-measures ANOVA showed significant differences across the three criteria (p = 0.014), and mixed-effects modeling demonstrated the same directional pattern. Treatment-focused questions received higher ratings (4.19 ± 0.24) than diagnostic questions (3.75 ± 0.31). Inter-rater reliability was moderate to good across domains. The intentionally incorrect response was correctly identified as inaccurate and/or unsafe by 20 of 22 raters (90.9%).
CONCLUSION: In this clinician-based expert review, ChatGPT® showed relatively strong performance in accuracy but lower performance in comprehensiveness and variable performance in safety when addressing FSD-related questions. These findings suggest that ChatGPT® may serve as a supplementary patient-education tool for basic, guideline-aligned topics, but its outputs should be regarded as informational starting points rather than stand-alone clinical recommendations.