作　　者： (）;

机构地区： 广州医科大学公共卫生学院统计学系

出　　处： 《中国医院统计》 2019年第3期165-169,共5页

摘　　要： 目的应用多种算法对社区卫生服务机构满意度进行综合评价,通过案例分析比较各种方法,为社区卫生服务机构满意度评价中的方法选择提供参考。方法以熵权法计算各指标客观权重,并在此基础上用线性求和法、灰色关联法和TOPSIS法从满意度不同维度上对案例的社区卫生服务机构进行排序及比较;训练一个具有高精度的满意度神经网络模型,以仿真值预测各维度的满意度评价结果并排序,并与传统算法比较。结果在传统方法中,线性加权法和灰色关联法对21家社区卫生服务机构4个维度评价中均为S17最优,S 2最差;在TOPSIS法中,方便可及维度最优为S17,最差为S9,另外3个维度均为S17最优,S2最差;3种传统方法结果相关性有统计学意义,相关系数为0.894~1.000之间,P<0.001。在BP神经网络中,4个维度最优均为S17,最差为S2,各维度仿真值误差分别为0.003、0.002、0.002、0.003;BP神经网络与3种传统方法相关性比较均有统计学意义,相关系数为0.891~1.000之间,P<0.001。结论各种方法均适用于社区卫生服务机构满意度评价,决策者可根据实际情况选择不同的方法对社区卫生服务机构的满意度进行评价。 Objective To comprehensively evaluate the satisfaction of community health services with different algorithms,to compare various methods through case analysis,and to provide methods for the satisfaction evaluation of community health services.Methods The entropy weight method was used to calculate the objective weight of each index,and on this basis,linear summation method,gray correlation method and TOPSIS method were used to rank and compare the differences among the community health services from different dimensions of satisfaction.A high precision neural network model of satisfaction was trained and the simulation values was used to predict satisfaction evaluation the results of each dimension.Then the results were sorted and compared with traditional algorithms.Results In the traditional methods,the linear weighting sum method and the gray correlation method were used to evaluate four dimensions of 21 community health service institutions with the best S17 and the worst S2.In the TOPSIS method,the best accessible dimension was S17,and the worst was S9.The other three dimensions were S17 best and S2 worst.The correlation between the results of the three traditional methods was statistically significant,and the correlation coefficient was between 0.894 and 1.000(P<0.001).In BP neural network,the optimal values of the four dimensions were S17 with the worst S2.The simulation errors of each dimension were 0.003,0.002,0.002 and 0.003.The correlation coefficient between BP neural network and the three traditional methods was between 0.891 and 1.000(P<0.001).Conclusion All the methods are applicable to the satisfaction evaluation of community health services,and decision-makers can choose different methods to evaluate the satisfaction of community health service institutions according to the actual situation.

关 键 词： 社区卫生服务 满意度 评价方法 比较研究

领　　域： []