The Research Of The Hospital Costs For Diabetes Patients Based On The Multilayer Perception Neural Network

ObjectiveTo analyze the composition of drug fees, examination fee, inspection fees, bedcosts and other expenses cost for patients with diabetes in the total cost. Build multilayerperception neural network model for the sub-costs and factors and analyze the influencingfactors on the sub-costs. Based on the composition of the sub-costs, we will measure influencedegree of various factors to the total hospital costs, so that we can draft the cost control measuresfor the decision makers in the hospital. The hospital fees can be better controlled and we canprovide the basis to improve the patients and hospital costs information system.MethodsCollecting the hospitalized information for845patients who were diagnoseddiabetes in a Class â…¢ Grade A hospital from Jan.2009to Dec.2011in xxx City. The statisticaldescription method is adopted to analyze the composition of the hospital costs and theinfluencing factors. Regard the sub-costs as the dependent variables; regard the gender, age,marital status, number of admissions, admission conditions, payment methods, length of stay,number of the days before surgery, treatment outcome, anesthesia and incision level as theindependent variables and then establish the multilayer perception neural network model forsub-costs and influencing factors. Analyze how important is the factors to the sub-costs. Basedon the sub-costs composition in the total hospital costs and how important are the influencingfactors to the sub-costs, weight how important are the influencing factors to the total cost ofhospitalization. We use SPSS17.0statistical software when doing the above analysis.ResultsIn the composition of the total hospital charges, drugs and inspection fees occupy thelargest proportion which respectively is40.7%and28.5%, both in the proportion of total hospitalcosts reached more than60percent; only drugs fee reaches nearly40percent. The greatestinfluential factor to the total hospital costs is the hospital stay (0.390), followed by paymentmethod (0.167), and the third is the treatment outcome (0.120), the fourth is the age of thepatients (0.103). The number of admissions (0.041) and the marital status (0.034) has theminimum impact on total hospital costs.ConclusionThrough this research, we can draw the following conclusions: to control thetotal hospital charges, we should mainly focus on hospital stay, payment method and treatmentoutcome. The result is satisfactory by using the multilayer perception neural network model toanalyze the influencing factors of the hospital costs. Based on the composition of the sub-costs and the importance of the factors, by weighting the importance of the influencing factors of thetotal hospital costs, we can get the result is reasonable.