潜在类别分析在缺血性中风病患者证候数据中的应用

The Application of Latent Class Analysis for Syndrome Data of Ischemia Stroke Patients

目的探讨观测结果为证候数据时潜在类别分析在样品聚类中的应用,及不同首发证候缺血性中风病患者各证候随时间的变化规律。方法采用潜在类别分析将1023例缺血性中风病住院患者按首发证候进行聚类,计算各组患者各时点各证候的频率,并绘制线图。结果根据潜在类别分析模型的拟合统计量及似然比检验得出4个潜在类别的模型是首选模型,缺血性中风病住院患者可聚为4个亚组,"内湿+血瘀"组98人,"痰+血瘀"组485人,"血瘀"组266人,"多种证候"组160人。经潜在类别分析聚成的4组患者,各证候的发生率及随时间的变化趋势不尽相同。结论潜在类别分析可用于观测结果为证候数据的样品聚类,聚类后各组各证候随时间的变化规律不同,对更深刻地揭示缺血性中风病的病机本质,更准确地指导临床中医药干预有实际意义。

Objective To explore the application of latent class analysis in sample cluster of syndrome data response variables, and syndrome variation regularity of ischemia stroke patients with different first syndrome. Methods Using latent class analysis,we gave 1023 patients a classification according to first syndrome, and calculated the syndrome fre- quency of different days, then drawed line graph. Results Four-class model is preferred according to fit statistics and likelihood ratio tests, and the four sub-groups number of patients was 98,485,266, and 160 respec- tively. The syndrome variation regularity of four sub-groups is different. Conclusion Latent class analysis can resolve the sample classification of syndrome data response variables. The difference in syndrome variation reg- ularity of sub-groups is more helpful to unveil the mechanism essence of is- chemia stroke, and guide clinical intervention.