基于临床路径的病例组合方法研究

目的:探讨基于临床路径的病例组合方法和人工神经网络在病例组合中的应用。方法:利用某综合性医院的一个临床路径流程下的5 2 3份出院病例资料,采用K -MODES聚类方法进行组合,用神经网络对预测病例的病例组合进行判断。结果5 2 3份病例聚为4组,各组间费用95 %可信区间互不重合;神经网络的训练误差为0 .0 0 2 9,病例组合预测和判断符合率为98.91%。结论:以临床路径下产生的病例为单元样本进行病例组合,结果更科学、客观。神经网络用于病例组合判断,不用确定单个节点变量的分割值,更符合病例组合由多变量共同作用的实情。

Study on the method of case-mix by appling clinical pathway

Objective:To investigate the methods of case-mix by clinical pathway:Methods:K-MEANS cluster analysis is applied to case-mix classification and artificial neural network is used for case-mix prediction.Results:523 inpatient records construct a case-mix classification scheme of 4 groups.Statistical significant difference of costs exists in 4 groups.The training error of artificial neural network is low(0.0029)and the predicting result is accurate(98.91%).Conclusion:The case-mix result is more reasonable using records under clinical pathway.The exsiting models of case-mix depend on dividing individual variables,but artificial neural network doesn't depend.