Study On Influencing Factors Of Coronary Heart Disease Medical Expenses Based On Multiple Statistical Methods

Objective: The rising cost of health care has brought a heavy financial burden.It is an effective way to make rational control of medical expenses by strengthening the analysis of medical expenses and finding the main influencing factors.However,different analytical tools produce different results and it is necessary to compare the differences among different analytical tools.Based on the analysis of medical expenses of hospitalized patients with coronary heart disease in ten counties of Liaoning Province,the distribution of hospitalization expenses was analyzed,and multiple linear regression,Logistic regression,BP neural network and support vector machine modeling were applied.The main influencing factors of hospital hospitalization cost were analyzed,and the most effective analysis model of the influencing factors of medical expenses was found.On this basis,we should rationally develop and utilize limited health resources to discuss the countermeasures for controlling the growth of medical expenses.Methods: The application conditions of data types and statistical models are studied through literature research.Taking the hospitalization information of coronary heart disease patients from 10 county medical institutions in Liaoning Province from January 2014 to December 2014 as an example,the K-means clustering method was used to group and discuss the CHD expense.Multiple linear regression,Logistic regression,BP neural network and support vector machine were used to analyze the influencing factors.According to the characteristics of medical expenses,the differences of the results of different models are found and the reasonable explanation is made.Result:1.A total of 4528 cases of hospitalized patients with coronary heart disease,including 2843 cases of female patients,accounting for 62.8%,1685 cases of male patients,accounting for 37.2%;the average age was 67.7±11.0years(25~95 years);59~73 years old age the largest,accounting for 49.01%;1~3 times accounted for 93.04% of the number of hospitalization;53.98% for emergency admission;the hospital for 4~9 days accounted for 67.51%;41.50% patients had complications;3.42% patients with complications.2.Among different hospitals,10 medical institutions in the D hospital most,1052 people(23.23%),at least 6 people hospital of G,(0.13%);the gender,in addition to F hospital(71)accounted for 46.41% of men,women accounted for 53.59%(82),the other 9 hospitals in the proportion of men from 31.54% to 39.95%,the proportion of women in 60.05% ~ 68.46%.The types of coronary heart disease,age group,group of hospitalization,admission,complication and complications were statistically different among different hospitals,P<0.001.3.The average cost of coronary heart disease was 5473.24 + 2839.60 yuan,the average daily cost was 659.62 + 266.96 yuan,and the average hospital days were 8.7 + 3.9 days.The hospitalization expenses of patients with different characteristics were statistically different.On the whole,the difference between the average hospital days,the average hospitalization expenses and the daily average cost between the ten hospitals was significant,and the difference was statistically significant.4.The cost of coronary heart disease in ten county hospitals is quite different.The average cost of H hospital is 6133.53 yuan,and the average cost of I hospital is 4599.02 yuan.The average cost of other seven hospitals is between 4623.56 and 5811.29,with a difference of 1187.73 yuan.However,the cost of each hospital in the high cost group is quite different.The average cost of the high cost group is between 7766.34 and 11940.81,with a difference of 4174.47 yuan.The cost of the low cost group is between 3593.90 and 5151.21,with a difference of 1557.31 yuan.5.The data are not suitable for multiple linear regression methods and can not be tested.The results of Logistic regression analysis showed that the influencing factors of hospitalization expenses included hospitalization days,medical institutions,times of hospitalization,type of coronary heart disease,gender and age: gender.The influencing factors of neural network models are: hospitalization days,types of coronary heart disease,medical institutions,hospitalization times,admission routes,complications,age,gender and complications.The factors of the support vector machine are the number of hospitalization,the type of coronary heart disease,the complication,the medical institution and the way of admission.And the accuracy of the target variables of SVM and ANN is 87.39% and 85.14% respectively.Conclusion: The distribution of patients' characteristics in ten hospitals was different.The average length of stay,average hospitalization expenses and daily average expenses between different medical institutions were statistically different.The three models of Logistic regression analysis and SVM and ANN were found to be related factors: the number of hospitalization days,type of coronary heart disease and medical institutions.Among them,the accuracy of the target variable obtained by SVM analysis is higher than that of the ANN model,and the number of variables selected by SVM is more reasonable.The influencing factors of SVM were in order of hospitalization days,type of coronary heart disease,complication,name of organization and route of admission.In this study,support vector machine has great advantages in the analysis of influencing factors of hospitalization expenses.There are no advantages and disadvantages in the model,but the applicable conditions of different models are different.Therefore,it is suggested that in the analysis of data,a variety of models can be used to analyze and select the most suitable model according to the characteristics of the data.