Journal francais d'ophtalmologie · 2025
Evaluation of appropriateness and readability of responses of large language models (ChatGPT-4o, Gemini, DeepSeek and Copilot) to frequently asked questions about congenital nasolacrimal duct obstruction.
СтатьяСравнительное исследованиеОценочное исследованиеНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Sönmez M M, Baydemir E E, Elibol E S
Аннотация
PURPOSE: To compare the appropriateness and readability of responses generated by four different large language models (LLMs) to frequently asked questions about congenital nasolacrimal duct obstruction (CNLDO).
METHODS: Twenty-five frequently asked questions (FAQs) related to CNLDO were answered by four LLMs: ChatGPT-4o, Gemini, DeepSeek, and Microsoft Copilot. Two experienced oculoplastic surgeons evaluated the responses for appropriateness and comprehensiveness using a Likert scale. In addition, the readability of the responses was assessed using an online readability tool.
RESULTS: ChatGPT-4o demonstrated significantly higher accuracy in the treatment and management category (P=0.001). Overall, DeepSeek provided the most appropriate responses across all categories (P=0.001). While DeepSeek achieved the highest total Likert scores, the differences among the models were not statistically significant (P=0.2). In terms of readability, Gemini and Copilot produced more readable texts, whereas DeepSeek's responses were more complex and structured.
CONCLUSIONS: DeepSeek showed superior performance in appropriateness and comprehensiveness across all questions related to CNLDO. However, its responses were less readable compared to those of the other LLMs, which may limit accessibility for some users.
METHODS: Twenty-five frequently asked questions (FAQs) related to CNLDO were answered by four LLMs: ChatGPT-4o, Gemini, DeepSeek, and Microsoft Copilot. Two experienced oculoplastic surgeons evaluated the responses for appropriateness and comprehensiveness using a Likert scale. In addition, the readability of the responses was assessed using an online readability tool.
RESULTS: ChatGPT-4o demonstrated significantly higher accuracy in the treatment and management category (P=0.001). Overall, DeepSeek provided the most appropriate responses across all categories (P=0.001). While DeepSeek achieved the highest total Likert scores, the differences among the models were not statistically significant (P=0.2). In terms of readability, Gemini and Copilot produced more readable texts, whereas DeepSeek's responses were more complex and structured.
CONCLUSIONS: DeepSeek showed superior performance in appropriateness and comprehensiveness across all questions related to CNLDO. However, its responses were less readable compared to those of the other LLMs, which may limit accessibility for some users.