B-P神经网络模型在缺血性心脏病住院费用影响因素分析中的应用

Application of B-P Neural Network Model in Analyzing the Influencing factors of Hospitalization Expenses of Ischemic Heart Disease

目的基于重庆市某三级甲等医院2016年1月1日-2016年12月31日的病案信息资料,探索B-P神经网络模型在缺血性心脏病住院费用影响因素分析中的应用。方法共纳入符合要求的缺血性心脏病病例1877例,其住院费用影响因素的分析采用B-P神经网络模型实现。结果建立的B-P神经网络模型训练集与测试集相对误差分别为0.331、0.336,住院天数、是否手术、缺血性心脏病类型、年龄、是否抢救为缺血性心脏病患者住院费用重要影响因素,其重要性分别为0.361、0.155、0.097、0.090、0.048。结论对于探索住院费用影响因素这类可能呈复杂非线性关系的问题,B-P神经网络是一种较优的建模方法。

Objectives To explore the application of B-P neural network model in analyzing the influencing factors of hospitalization expenses of ischemic heart disease, based on the medical record information of a Three-A level hospital in Chongqing in 2016. Methods A total of 1877 patients with ischemic heart disease were included in the study. The analysis of the influencing factors of hospitalization expenses was carried out by B-P neural network model. Results The relative error of training set and test set were 0.331, 0.336 respectively. Hospital stay,decision to operate, types of ischemic heart disease, age, rescue situation have important influence on hospitalization expenses of ischemic heart disease, the importance of which were 0.361, 0.155, 0.097, 0.090,0.048 respectively. Conclusions To explore the influence factors on admission expenditures of this kind disease may be complex nonlinear relationship problems, B-P neural network model is a preferred modeling method.