Oral diseases · 2026
Quality and Readability of Large Language Models' Responses to Oral Lichen Planus Patients' FAQs.
СтатьяНазвание 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.
Polizzi Alessandro, Isola Gaetano, Caponio Vito Carlo Alberto, Santos-Silva Alan Roger, González-Serrano José, Albuquerque Rui, Brailo Vlaho, Farag Arwa Mohammad, López Jornet María Pía, Robledo Sierra Jairo, Sollecito Thomas Peter, Dan Hongxia, Diniz Freitas Márcio, Bissonnette Caroline, Wiriyakijja Paswach, Hernández Gonzalo, López-Pintor Rosa María
Аннотация
OBJECTIVE: To evaluate the quality and readability of large language models (LLMs) when responding to Frequently Asked Questions (FAQs) about oral lichen planus (OLP).
METHODS: We evaluated the responses of three LLMs (ChatGPT-4o, Gemini 2.0 Flash Experimental, and Copilot) to 13 patient-centered FAQs about OLP. Questions were identified using query tools, and answers were assessed by 14 oral medicine experts using the Quality Assessment of Medical Artificial Intelligence (QAMAI) tool. Readability was analyzed with the Flesch Reading Ease (FRE) and Flesch-Kincaid Grade Level (FKG) tools.
RESULTS: All LLMs provided generally accurate and relevant responses, with median QAMAI scores indicating "good" to "very good" quality. ChatGPT achieved slightly higher completeness, particularly for questions on OLP definition and treatment. The reference provision was inconsistent across all chatbots. Readability analysis revealed that most responses required college-level literacy, with ChatGPT producing the most complex texts, Gemini occasionally achieving more accessible outputs, and Copilot situated in an intermediate position.
CONCLUSIONS: LLMs may have potential as adjunctive tools for patient education in OLP, although they remain limited by incomplete information, inconsistent references, and suboptimal readability. Future research should incorporate longitudinal LLMs evaluations and training to develop models delivering accurate, accessible information, tailored to users' literacy levels.
METHODS: We evaluated the responses of three LLMs (ChatGPT-4o, Gemini 2.0 Flash Experimental, and Copilot) to 13 patient-centered FAQs about OLP. Questions were identified using query tools, and answers were assessed by 14 oral medicine experts using the Quality Assessment of Medical Artificial Intelligence (QAMAI) tool. Readability was analyzed with the Flesch Reading Ease (FRE) and Flesch-Kincaid Grade Level (FKG) tools.
RESULTS: All LLMs provided generally accurate and relevant responses, with median QAMAI scores indicating "good" to "very good" quality. ChatGPT achieved slightly higher completeness, particularly for questions on OLP definition and treatment. The reference provision was inconsistent across all chatbots. Readability analysis revealed that most responses required college-level literacy, with ChatGPT producing the most complex texts, Gemini occasionally achieving more accessible outputs, and Copilot situated in an intermediate position.
CONCLUSIONS: LLMs may have potential as adjunctive tools for patient education in OLP, although they remain limited by incomplete information, inconsistent references, and suboptimal readability. Future research should incorporate longitudinal LLMs evaluations and training to develop models delivering accurate, accessible information, tailored to users' literacy levels.