Journal of medical education and curricular development · 2026
A Pilot Study of a CustomGPT for the Royal College of Ophthalmologists Curriculum 2024.
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.
Purcell William, Chadha Vikas, Gogi Devina, Long Vernon, Shafi Aabgina, Ramkissoon Yashin
Аннотация
BACKGROUND: Large language models (LLMs) employ transformer architectures to generate contextually appropriate responses and are increasingly applied in medical education. To support implementation of the Royal College of Ophthalmologists (RCOphth) Ophthalmic Specialist Training (OST) Curriculum 2024, the Yorkshire & Humber School of Ophthalmology developed a CustomGPT trained exclusively on official RCOphth curriculum documents. This study evaluated its usability, effectiveness, and accuracy.
METHODS: A pilot study invited ophthalmology trainees (ST1-ST7, Fellows) and trainers (Consultants, Supervisors, Tutors, Programme Directors) to use the CustomGPT over a three-month period (April to June 2025). Quantitative feedback (Likert scales) assessed relevance, ease of use, and satisfaction, while qualitative feedback explored perceived value and limitations. Accuracy was evaluated by comparing responses from the CustomGPT and the RCOphth Training Committee (TC) Chair to ten authentic curriculum queries.
RESULTS: Twenty-eight participants (14 trainees, 14 trainers) completed the survey; only 21% were regular AI users. Mean scores were high for relevance (4.21/5), ease of use (4.38/5), and satisfaction (4.10/5), with 86% reporting time savings and 75% rating their likelihood of recommending the tool as 8/10 or higher. Qualitative data highlighted concise, accurate, and well-structured responses but noted occasional technical issues and rigid interpretation of complex queries. Comparative analysis showed strong alignment with the TC Chair, though the GPT sometimes demonstrated greater precision in referencing but less flexibility in interpretation.
CONCLUSIONS: A curriculum-specific GPT can provide rapid, accurate guidance, reducing ambiguity and improving efficiency. When integrated into training platforms, such tools can enhance curriculum accessibility in postgraduate medical education.
METHODS: A pilot study invited ophthalmology trainees (ST1-ST7, Fellows) and trainers (Consultants, Supervisors, Tutors, Programme Directors) to use the CustomGPT over a three-month period (April to June 2025). Quantitative feedback (Likert scales) assessed relevance, ease of use, and satisfaction, while qualitative feedback explored perceived value and limitations. Accuracy was evaluated by comparing responses from the CustomGPT and the RCOphth Training Committee (TC) Chair to ten authentic curriculum queries.
RESULTS: Twenty-eight participants (14 trainees, 14 trainers) completed the survey; only 21% were regular AI users. Mean scores were high for relevance (4.21/5), ease of use (4.38/5), and satisfaction (4.10/5), with 86% reporting time savings and 75% rating their likelihood of recommending the tool as 8/10 or higher. Qualitative data highlighted concise, accurate, and well-structured responses but noted occasional technical issues and rigid interpretation of complex queries. Comparative analysis showed strong alignment with the TC Chair, though the GPT sometimes demonstrated greater precision in referencing but less flexibility in interpretation.
CONCLUSIONS: A curriculum-specific GPT can provide rapid, accurate guidance, reducing ambiguity and improving efficiency. When integrated into training platforms, such tools can enhance curriculum accessibility in postgraduate medical education.