A new analysis published in Frontiers in Radiology reveals the potential of large language models (LLMs) to assist in patient communication regarding interventional radiology procedures. The study compared responses from LLMs, including ChatGPT-4o, Gemini Pro, and Microsoft Copilot, to those of a human expert regarding patient queries on varicocele embolization. The results demonstrated that LLMs not only provided accurate answers but also exhibited empathy, outperforming the human expert in this study. Gemini achieved the highest scores for accuracy and empathy, followed closely by Copilot and ChatGPT. The authors suggest that AI could enhance patient education in interventional radiology, although they caution that LLMs require ongoing monitoring for accurate patient information.
Thu, 02 Oct 2025 19:32:48 GMT | Radiology Business