Surgery on Children
The potential of AI applications to providing clinical counseling to parents: a pilot study
By Nilsun Kuas, Ozgur Aktas, Ergün Bahadırhan Karkın and 1 others.
Abstract
Aims
This study aimed to evaluate the responses of current large language model-based AI applications to pediatric surgery and urology clinical scenarios using expert physician-defined criteria, and to investigate their potential for providing clinical guidance to parents.
Methods
Ten clinical questions were submitted to three AI applications (ChatGPT-5.2, DeepSeek-V3.1, Gemini 3). Responses were scored blindly by four specialist physicians using ten criteria across four domains (maximum, 20 points per question). A safety veto criterion (E1) was included as an independent safeguard; any response failing to recognize a clinical emergency or containing potentially harmful information was rendered invalid regardless of other scores.