应用神经网络方法对妇科手术病例DRGs分组测算住院费用  

Application of neural networks to measuring the hospitalization expenses on cases of gynecological surgeries based on DRGs

作　　者：方良欣[1] 邓庆珊[1] 梁燕霞[1] 彭嘉怡[1] 邓群娣[1] 

机构地区：[1]广东省妇幼保健院病案科,广东广州510010

出　　处：《华夏医学》2013年第3期518-522,共5页Acta Medicinae Sinica

基　　金：广东省医学科学技术研究基金申请项目资助(A2011080)

摘　　要：目的:基于DRGs分组,尝试将相关疾病属性采用神经网络方法建立数学模型,以此估算住院费用的可信度。方法:回顾性分析本院2010～2012年妇科各类疾病并以手术治疗的病例,选取某些疾病相关属性应用神经网络建立数学模型进行测算费用。将实际费用和测算费用作比较。结果:使用配对设计的秩和检验实际和测算费用,Z=0.806,P=0.420>0.05,两者没有统计学差异。结论:应用神经网络模型测算费用与实际费用接近,可作为支付参考。Objective： To test the reliability of assessing hopitalization expenses by a mathematical model established by applying neural network to analyzing the related attributes of diseases based on diagnosis-related groups （DRGs）. Methods： Gynecological cases with operational treatment from 2010 to 2012 in our hospital were retrospectively analyzed. The hospitalization expenses of the related attributes of some diseases were culculated by the mathematical model estabished by the neural network. Then the the actual costs and calculated costs were compared. Results： Paire-de- signed rank-sum test was applied to testing the actual costs and the calculated costs （Z^0.806 ,P~ O. 420~,0. 05）, there is no significance between the actual costs and the measured costs. Conclusion： The expenses calculated by the mathematical model is close to the actual expenses, so it can be used as a payment reference.

关 键 词：疾病诊断相关分类系统(DRGs) 相关属性 神经网络 

分 类 号：R197.323[医药卫生—卫生事业管理]