Back to the list
Congress: ECR25
Poster Number: C-28535
Type: Poster: EPOS Radiologist (scientific)
Authorblock: S. Jankowski1, D. C. Rotzinger1, F. Poncet1, M. Nowak1, M. Gulizia1, J. Richiardi1, F. Ria2, C. Mourad3, C. Pozzessere1; 1Lausanne/CH, 2Durham, NC/US, 3Beyrouth/LB
Disclosures:
Sofyan Jankowski: Nothing to disclose
David Christian Rotzinger: Nothing to disclose
Florian Poncet: Nothing to disclose
Marie Nowak: Nothing to disclose
Marianna Gulizia: Nothing to disclose
Jonas Richiardi: Nothing to disclose
Francesco Ria: Nothing to disclose
Charbel Mourad: Nothing to disclose
Chiara Pozzessere: Nothing to disclose
Keywords: Artificial Intelligence, Catheter arteriography, CT, MR, Diagnostic procedure, Radiation safety, Technology assessment, Patterns of Care
Conclusion

Unlike experts, who rated ChatGPT comparably to official institutional websites in terms of scientific adequacy, public comprehension, and overall satisfaction, patients preferred the AI-generated texts for understandability, trust, reassurance, and satisfaction. Interestingly, patients were unable to distinguish between ChatGPT and human responses. These findings suggest that while the radiation protection information provided by institutional websites is generally well-received by experts, it remains imperfect for patients. Conversely, ChatGPT appears more suitable for providing radiation protection information to the public and may be a valuable tool in assisting health professionals with patient communication. However, this should be considered in light of other studies that highlight patients' low trust in healthcare systems' ability to use AI responsibly and protect them from potential AI-related harms.(9)

GALLERY