大规模中医师考试评价中认知诊断模型的构建及应用

Analysis and Assessment of a Large-Scale Traditional Chinese Medicine Examination Based on Cognitive Diagnostic

目的采用认知诊断模型对中医师岗位胜任力第二阶梯考核理论考试进行分析,为大规模医学考试的认知诊断研究提供参考。方法选择2023年参加中医师岗位胜任力第二阶梯考核理论考试的16310位考生作答情况开展认知诊断分析。检验拓广决定型输入噪声与门模型(G-DINA模型)和考试数据之间的拟合程度;在认知诊断框架下进行项目参数分析和信度分析;从不同层面综合分析考生的认知属性掌握情况。结果本研究中16310名考生平均分数为(189.76±40.86)分,共有9994名考生通过了此次考试,通过率为61.28%。考试总题量300题,五个模块中中医基础考核频率最高(132题,占44%),其次是中医临床(87题,占29%)、西医基础(60题,占20%),中医经典(12题,占4%)和医学人文(9题,占3%)考核比例较少。模型拟合指标结果显示,G-DINA模型曲线下面积(AUC)为0.98,说明该模型与本研究数据拟合优秀。项目参数分析结果显示,300道测试题目认知诊断模型的所有区分度均大于0,平均区分度为0.25,整体题目区分度的表现良好。信度分析结果显示,测验水平分类准确率为0.75,属性水平分类准确率为0.83~0.99,表明G-DINA模型的属性分类准确度高。认知诊断分析结果显示,相较于未通过考试的考生,已通过考试的考生对于各模块认知属性的掌握比例更高。结论基于认知诊断的大规模中医考试分析,可为提高中医学人才教育和培养质量提供支撑。

Objective To analyze the scores of the second-tier assessment on theory of competency of Traditional Chinese Medicine(TCM)physicians based on cognitive diagnostic model,and to provide a reference for cognitive diagnostic studies of large-scale medical examinations.Methods Cognitive diagnostic analyses were conducted by selecting the responses of 16,310 candidates who took the theory examination of the second-tier assessment on the competency of TCM physicians in 2023.The degree of fit between the topological deterministic input noise and gate model(G-DINA model)and the examination data was examined;item parameter analysis and reliability analysis were conducted under the cognitive diagnostic framework;the candidates'mastery of cognitive attributes was comprehensively analysed at different levels.Results The mean score of 16,310 candidates in this study was(189.76±40.86),and a total of 9,994 candidates passed this examination,with a pass rate of 61.28%.The total number of questions in the examination was 300,and among the five modules,the frequency of assessment of TCM fundamentals was the highest(132 questions,44%),followed by TCM clinic(87 questions,29%),Western medicine fundamentals(60questions,20%),and TCM classics(12 questions,4%)and medical humanities(9 questions,3%)was less.The results of the model fit index showed that the area under the curve(AUC)of the G-DINA model was 0.98,indicating that the model fit the study data excellently.The results of item parameter analysis showed that all the discriminations of the cognitive diagnostic model for the 300 test questions were greater than 0,with an average discrimination of 0.25,and the overall performance of the question discriminations was good.The results of the reliability analysis showed that the test level classification accuracy was 0.75,and the attribute level classification accuracy was 0.83 to 0.99,indicating that the G-DINA model has high attribute classification accuracy.The results of the cognitive diagnostic analyses showed that the proportion of candidates who had passed the exam was higher for the cognitive attributes of each module compared to those who had not passed the exam.Conclusion The analysis of large-scale TCM examinations based on cognitive diagnosis can provide support for improving the quality of TCM talent education and cultivation.