Dataset of clinical cases, AI response and experts response from the research: Antimicrobial Stewardship in the Era of AI: A Head-to-Head Comparison of a Machine Learning HTL Algorithm and Large Language Models in Real-World Infectious Disease Cases

The following dataset contains information used for the study titled: Antimicrobial Stewardship in the Era of AI: A Head-to-Head Comparison of a Machine Learning HTL Algorithm and Large Language Models in Real-World Infectious Disease Cases. This dataset contains 88 clinical cases of urinary tract infections extracted from anonymized medical records (to exclude patients’ personal data). It also includes portions of the antibiotic susceptibility test reports from each record. Based on this data, prompts were created and presented to the artificial intelligence models for analysis, with the aim of obtaining recommendations for initial and secondary antimicrobial treatment. We examined whether there was agreement or disagreement among the models’ responses, as well as the type of disagreement analyzed. Finally, we present the categorization of the responses provided by the three infectious disease experts, compared to the models’ responses.

本数据集用于支撑题为《人工智能时代的抗菌药物管理：机器学习HTL算法与大语言模型（Large Language Model）在真实世界感染病病例中的头对头比较》的研究。本数据集包含从匿名化病历中提取的88例尿路感染临床病例（已剔除患者个人隐私信息），同时涵盖每份病历中的抗生素敏感性试验报告节选内容。基于上述数据，研究人员构建了提示词并提交至各人工智能模型进行分析，旨在获取初始与次级抗菌治疗方案建议。本研究评估了各模型回复间的一致性与分歧情况，并对分歧类型展开分析。最后，本文呈现了3名感染病专科专家对模型回复的分类结果，并与模型输出进行对比。