Patient Perceptions of Artificial Intelligence-Generated Kidney Transplant Information: Comparing ChatGPT With the National Kidney Foundation

Abstract

Rationale & Objective: Generative artificial intelligence (AI) may help patients better understand the complexities of kidney transplantation. However, little is known about how individuals with chronic kidney disease (CKD) perceive AI-generated health information. This study assessed patient perceptions of AI-generated responses to common kidney transplant queries compared to those from a trusted health resource. Study Design: A cross-sectional online survey. Setting & Participants: A total of 216 adults with CKD, including kidney transplant recipients, residing in the United States participated in the study. Exposures: Participants compared kidney transplant-related query responses generated by ChatGPT (GPT-4o), a widely used generative AI tool, with those provided by the National Kidney Foundation (NKF). Outcomes: Participant perceptions across several domains: overall preference, perceived information quality, empathy, and learning outcomes. Analytical Approach: Participants reviewed paired responses from both ChatGPT and NKF, presented without source attribution. Results were analyzed using mixed-effect models. Results: < 0.001). Limitations: The web-based survey may not fully represent the diverse populations served by transplant centers. Limited prompts were used, which may not capture the full range of transplant scenarios. We were also unable to determine which specific features influenced participant preferences. Conclusions: Generative AI platforms like ChatGPT may present information in ways that resonate with patients, potentially enhancing their education and engagement. However, as these tools are still in the early stages of integration into everyday life, their use should be guided by careful human oversight.

Ähnliche Arbeiten