Analysis And Evaluation On Health Resource Allocation Efficiency In China With Network DEA And BP ANN

Objective: To propose a method based on the combination of network data envelopment analysis and artificial neural network to evaluate the efficiency of health resource allocation in China from 2012 to 2016.Aim to provide scientific suggestions for rational allocation of health resources by discussing the difference and its reason in the efficiency of regional health resource allocation.The specific objectives are as followed:(1)summary two common models used for evaluating the efficiency of health resource allocation,explore its relevant theory,evolution,advantages and disadvantages.(2)Construct a network data envelope model to evaluate the efficiency of health resource allocation in 31 provinces of China from 2012 to 2016 in terms of time series,and analyze the improvement and target values of each index in inefficient province;(3)Construct an artificial neural network model,combining the evaluation results of the network data envelopment analysis model,to comprehensively analyze the efficiency of health resource allocation in 31 provinces of China in 2016;(4)Provide theoretical basis for optimizing the allocation of regional health resources.Content: Network DEA model were applied to analyze the efficiency of health resource allocation in 31 provinces of China during 2012 to 2016,including overall efficiency and production efficiency of healthy human resources,healthy material resources and medical services.And investigate the reasons of inefficiency and put forward specific improvement of relevant target.Secondly,the results of efficiency evaluation using the Network DEA model are included as expected outputs into the BP artificial neural network model.To ultimately analyze the date of 2016,the data from 2012 to 2014 and the date of 2015 were used as training sample and test sample respectively.And analyze the current situation of health resource allocation and the regional differences in terms of cross-sectional analysis.Finally,summary the meliority in identity with the combination of Network DEA model and BP artificial neural network,and provide reference for establishing a relatively reliable efficiency evaluation system of health resource allocationMethods:(1)Network DEA: According to the principle of index selection,an index system was established to evaluate the efficiency of overall health resource allocation and each node during the study.(2)BP artificial neural network: Select more comprehensive index as the ingredient of input layer of neural network,the evaluation results using Network DEA were set as output layer,BP artificial neural network was employed in training,testing and identifying the efficiency of health resource allocation,and correct the efficiency value acquired with network DEA model according to the result of final identification.Aim to evaluate the efficiency of our health resource allocation in manner of comprehensive,multi-dimensional ways.Results: for overall efficiency,the proportion of continuously ineffective provinces from 2012 to 2016 was 93.5%,the average efficiencies of the eastern,central and western regions were 0.581,0.705,and 0.682 respectively.By decomposing the overall efficiency into production efficiency of healthy human resources,healthy material resources and medical services,from 2012 to 2016,the proportion of provinces whose efficiency value was 1 were 12.9%,19.4%,16.1%,16.1%,61.3% at node 1,and were 32.3%,41.9%,32.3%,38.7%,12.9% at node 2,respectively.The efficiency of node 1 was higher than that of node 2 in 16(51.6%)regions.The analysis of the improvement and target values of each index in inefficient province detected that the allocation of healthcare settings,number of beds,doctors,and nurses are over or under-satisfied,and the amount of change varies;however,there are varying degrees of deficiencies in the output of medical services as the representative of number of visits,out of hospital and physical examination.The results of the artificial neural network model detected that the average efficiency of health resource allocation in 31 provinces of China was 0.638 in 2016,and inefficient province occupied 96.8%.Among the inefficient province,the efficiencies of Shanxi,Shandong,Henan,Guangxi province were higher than 0.8.There are 7(22.6%)regions with efficiency values lower than 0.5,among which 4 are located in the eastern region and 3 are located in the western region.Conclusion:(1)From 2012 to 2016,in addition to individual province,the efficiency of health resource allocation in China generally showed a downward trend,and there are varying degrees of differences in efficiency values of regions at different stage.(2)During the current study,the efficiency of converting the total health expenditure into health human resource and material resource was significantly improved in all regions,while the efficiency of converting health human resource and material resource into medical services was little reduced.The health resource in lots of regions were lack of efficient allocation.(3)When the indicators such as economy,population,and healthy level of residents were introduced into the artificial neural network model,the efficiency of resource allocation in most regions need reviewed.The average efficiency value of health resource allocation in the central region was the highest,followed by western region and eastern region.And the allocation efficiency of health resources in Beijing,Tianjin,and Shanghai was low.(4)The combination of Network DEA and BP artificial neural network not only makes up for the deficiency of identifying,but also improves the accuracy and precision of the efficiency measurement and realizes the improvement of the original model.