目的：探索应用单维项目反应模型与经典测验理论相结合的方法，解决量表评价问题。方法：对841例亚健康二手数据实证分析，进行信度、效度以及项目角度的测量学评价。结果：5个领域的克朗巴赫a系数均高于0.6，内部一致性较高；5个领域的量表结构效度较好，CFI为0.96；项目区分度>0.3，项目对所属领域诊断的灵敏度较高；量表信息量为33.8，对亚健康状态估计精度较高；但仍有部分条目需要进一步修订。结论：单维项目反应模型通过单维性、局部独立性检验，可以用来评价项目的心理学测量性能，对经典测验理论评价结果进行补充。亚健康量表信、效度较高，经IRT校正后的项目可以用于开展计算机自适应测试，进行人机交互“治未病”动态监测。

This study was aimed to assess scales with the unidimensional item response models (UIRM) and classical test theory. The reliability and validity of 841 sub-health status samples were analyzed by CTT. Surveying evaluation from the project perspective was also conducted. The results showed that Cronbach’s α coefficients in five areas were all larger than 0.6 with relatively high consistency. The scale structures of five areas were relatively good and the CFI was 0.96. When discriminations of all items were larger than 0.3, the sensitivity of item for its area was high. When the scale test information was 33.8, the accuracy of sub-health status was relatively high. However, there were still some items should be further revised. It was concluded that UIRM can be used to evaluate the psychological measurement performance in the project. It is a supplement to the evaluation result of classical test theory. The reliability and validity of sub-health status scale are relatively high. After IRT correction, it can be used for human-computer interaction techniques“preventive treatment”dynamic monitoring.

中国中医科学院第九批自助选题（Z0409）：基于多维IRT的健康状态评估系统构建研究，负责人：白文静。