Photo Credit: PeopleImages

The following is a summary of “Evaluating the Utility of ChatGPT in Enhancing Parental Education and Clinical Support in Hypospadias Care,” published in the January 2025 Journal of Pediatric Neurology by Adriansyah et al. 

Hypospadias is a common congenital anomaly requiring timely parental education to ensure informed decision-making regarding diagnosis, treatment, and postoperative care. However, many existing online resources exceed recommended readability levels, limiting accessibility for non-medical audiences. Artificial intelligence-driven tools like ChatGPT offer a potential solution by generating conversational and easily digestible medical content. This study evaluates the utility of ChatGPT-4 in providing clear, accurate, and actionable information for parental education on hypospadias. 

A structured set of questions covering key aspects of hypospadias, including diagnosis, treatment options, and postoperative care, was posed to ChatGPT-4. The responses were quantitatively assessed using the Patient Education Material Assessment Tool for Printable Materials (PEMAT-P), which measures understandability and actionability. Additionally, six pediatric urologists conducted a qualitative evaluation of the generated information, rating accuracy on a four-point scale (1 = completely accurate, 4 = completely inaccurate). Inter-observer agreement among the urologists was determined using Fleiss’ Kappa statistic. The quantitative analysis demonstrated high understandability, with scores ranging from 84% to 92% (mean: 88%), indicating that the AI-generated content was largely comprehensible to a lay audience. However, actionability scores varied significantly, ranging from 37% to 70% (mean: 51%), highlighting the AI’s limited ability to provide clear, actionable steps for parents.

In the qualitative assessment, 41% of ChatGPT responses were rated as completely accurate, 35% as accurate but lacking essential details, and 24% as inaccurate. The Fleiss’ Kappa value of 0.607 indicated substantial agreement among reviewers regarding response accuracy. These findings suggest that while ChatGPT-4 can effectively generate understandable content on hypospadias, significant improvements are needed to enhance the actionability and reliability of its information. AI-generated content may still introduce inaccuracies, which could misguide parents seeking medical advice. Future iterations of AI-driven educational tools should incorporate improved validation mechanisms, real-time updates from evidence-based medical guidelines, and the integration of visual aids to enhance comprehension. Further research is warranted to explore the role of AI in supplementing physician-led parental education while addressing its limitations to ensure safe and effective use in clinical settings.

Source: jpurol.com/article/S1477-5131(25)00028-2/fulltext