Adriansyah Ihsan Azka, Wahyudi Irfan, Vallasciani Santiago, Elhout Yaser, Bhat Amilal, Priyadarshi Shivam, Ansari M S, Raharja Putu Angga Risky, Abbas Tariq O
Department of Urology, Faculty of Medicine, Universitas Indonesia, Jakarta, Indonesia.
Urology Division, Sidra Medicine, Doha, Qatar.
J Pediatr Urol. 2025 Jun;21(3):639-643. doi: 10.1016/j.jpurol.2025.01.026. Epub 2025 Jan 31.
Hypospadias is a prevalent congenital anomaly that requires effective parental education. Current online resources often exceed recommended readability levels, potentially hindering understanding. This study evaluates the utility of ChatGPT in providing accurate, clear, and actionable information towards parental education about hypospadias.
A structured set of questions was posed to ChatGPT 4.0 covering diagnosis, treatment options, and postoperative care. Responses were quantitatively assessed using the Patient Education Material Assessment Tool for Printable Materials (PEMAT-P) to measure understandability and actionability. Qualitative evaluations were conducted by six pediatric urologists who rated the information for accuracy on a scale from 1 (completely accurate) to 4 (completely inaccurate). The Fleiss' Kappa statistic was calculated to assess inter-observer agreement among the urologists.
The quantitative assessment yielded understandability scores between 84 % and 92 % (average 88 %), while actionability scores ranged from 37 % to 70 % (average 51 %). In the qualitative assessment, 41 % of responses were deemed completely accurate, with 35 % considered accurate but inadequate, and 24 % rated as inaccurate. The overall Kappa value was 0.607, indicating substantial agreement among reviewers regarding the accuracy of the information provided by ChatGPT.
ChatGPT can effectively convey information about hypospadias, but enhancing the actionability of its responses is crucial. Inaccuracy is still a main concern in using AI-generated search engine. Future updates should include more accurate and reliable responses and visual aids addition may support parents in navigating their child's care.
尿道下裂是一种常见的先天性异常,需要对家长进行有效的教育。当前的在线资源往往超出了推荐的可读性水平,可能会妨碍理解。本研究评估了ChatGPT在为家长提供关于尿道下裂的准确、清晰且可行的信息方面的效用。
向ChatGPT 4.0提出了一组结构化问题,涵盖诊断、治疗选择和术后护理。使用可打印材料的患者教育材料评估工具(PEMAT-P)对回答进行定量评估,以衡量可理解性和可行性。由六位小儿泌尿科医生进行定性评估,他们以1(完全准确)至4(完全不准确)的量表对信息的准确性进行评分。计算Fleiss' Kappa统计量以评估泌尿科医生之间的观察者间一致性。
定量评估得出的可理解性分数在84%至92%之间(平均88%),而可行性分数在37%至70%之间(平均51%)。在定性评估中,41%的回答被认为完全准确,35%被认为准确但不充分,24%被评为不准确。总体Kappa值为0.607,表明评审者在ChatGPT提供的信息准确性方面存在实质性一致。
ChatGPT可以有效地传达关于尿道下裂的信息,但提高其回答的可行性至关重要。不准确仍然是使用人工智能生成的搜索引擎的一个主要问题。未来的更新应包括更准确可靠的回答,添加视觉辅助工具可能有助于家长了解孩子的护理情况。