Driesnack Sebastian, Rücker Fabian, Dietze-Jergus Nadine, Bondarenko Alexander, Pletz Mathias W, Viehweger Adrian
Institute for Medical Microbiology and Virology, University of Leipzig Medical Center, Liebigstraße 21, 04103 Leipzig, Germany.
Institute for Infectious Diseases and Infection Control, Jena University Hospital, Am Klinikum 1, 07747 Jena, Germany.
JAC Antimicrob Resist. 2024 Jul 6;6(4):dlae099. doi: 10.1093/jacamr/dlae099. eCollection 2024 Aug.
Scalable teaching through apps and artificial intelligence (AI) is of rising interest in academic practice. We focused on how medical students could benefit from this trend in learning antibiotic stewardship (ABS). Our study evaluated the impact of gamified learning on factual knowledge and uncertainty in antibiotic prescription. We also assessed an opportunity for AI-empowered evaluation of freeform answers.
We offered four short courses focusing on ABS, with 46 participating medical students who self-selected themselves into the elective course. Course size was limited by the faculty. At the start of the course, students were given a questionnaire about microbiology, infectious diseases, pharmacy and qualitative questions regarding their proficiency of selecting antibiotics for therapy. Students were followed up with the same questionnaire for up to 12 months. We selected popular game mechanics with commonly known rules for teaching and an AI for evaluating freeform questions.
The number of correctly answered questions improved significantly for three topics asked in the introductory examination, as did the self-assessed safety of prescribing antibiotics. The AI-based review of freeform answers was found to be capable of revealing students' learning gaps and identifying topics in which students needed further teaching.
We showed how an interdisciplinary short course on ABS featuring gamified learning and AI could substantially improve learning. Even though large language models are a relatively new technology that sometimes fails to produce the anticipated results, they are a possible first step in scaling a tutor-based teaching approach in ABS.
通过应用程序和人工智能（AI）进行可扩展教学在学术实践中越来越受到关注。我们关注的是医学生如何能从抗生素管理（ABS）学习的这一趋势中受益。我们的研究评估了游戏化学习对抗生素处方方面事实性知识和不确定性的影响。我们还评估了利用人工智能对自由形式答案进行评估的机会。
我们提供了四门聚焦于ABS的短期课程，46名医学生参与其中，他们自行选择参加这门选修课程。课程规模受教师限制。在课程开始时，给学生发放了一份关于微生物学、传染病、药学以及关于他们选择抗生素进行治疗的熟练程度的定性问题的问卷。对学生进行长达12个月的随访，使用相同的问卷。我们选择了具有常见规则的流行游戏机制用于教学，并使用人工智能评估自由形式的问题。
入门考试中所问的三个主题的正确回答问题数量显著增加，开具抗生素的自我评估安全性也有所提高。基于人工智能的自由形式答案审查能够揭示学生的学习差距，并确定学生需要进一步教学的主题。
我们展示了一门以游戏化学习和人工智能为特色的跨学科ABS短期课程如何能大幅提高学习效果。尽管大语言模型是一项相对较新的技术，有时无法产生预期的结果，但它们是在ABS中扩展基于导师的教学方法的可能第一步。
