Appropriateness of Prostate Cancer Prevention and Screening Recommendations Obtained From ChatGPT-4

Abstract

<title>Abstract</title> <bold>Purpose:</bold> To evaluate the appropriateness of prostate cancer (PCa) prevention and screening recommendations provided by ChatGPT-4 <bold>Materials and Methods:</bold> We presented 32 questions regarding PCa prevention and screening using ChatGPT-4 based on National Comprehensive Cancer Network (NCCN) guidelines and clinical experience. Each question was asked to ChatGPT-4 three times, and the responses were graded by three expert urologists as reviewers. We evaluated sets of three responses generated by ChatGPT-4 in two distinct contexts: a general patient-facing platform and direct patient-to-clinician messages. A board certified/eligible urologist assessed each response set as appropriate, inappropriate, or unreliable. <bold>Results:</bold> In both contexts, ChatGPT-4 generated appropriate responses for 30 out of 32 (94%) questions, while one (3%) was considered inappropriate and one (3%) was deemed unreliable. Inappropriate response was related to a Gleason score of 5+3, while unreliable answer was linked to a Gleason score of 3+5 on prostate biopsy results. <bold>Conclusions:</bold> ChatGPT-4 is capable of providing appropriate counseling regarding disease prevention and screening for patients with PCa.

Ähnliche Arbeiten