机构地区: 内江师范学院计算机科学学院,四川内江641199
出 处: 《内江师范学院学报》 2017年第8期69-72,共4页
摘 要: 为解决医保基金被滥用、被浪费的问题,本文对某地区医保中心的数据进行分析,将问题归结为找出违规可能性较大的"可疑"处方,排除大部分正常处方,以达到减少人工审核量的目的.运用离群点检测方法,在对医保数据进行预处理分析、解决其高维稀疏问题的基础上,提出了一种面向医保审核的属性权重计算公式,以提高检测的准确率;将KNN算法应用于白内障、胆结石、阑尾炎三种疾病的病例处方检测.实验显示,KNN离群点检测算法能检测出大部分的"可疑"处方. To solve the abuse and squandering of medical insurance fund,through the analysis of the data from a local medical insurance center,a simplified method,in which the major task of verification is to screen out those"suspicious"medical prescriptions and let go of those normal ones,was found to reduce the workload of human verification.By use of the outlier detection algorithm and on the basis of solving the high dimensional sparsity after a pre-processing analysis of the medical insurance data,a computational formula of attribute weights is worked out for medical insurance verification for the purpose of improving the detection accuracy;by applying the KNN algorithm to the detection of prescriptions of patients afflicted with cataract,gall-stone and appendicitis,it is found that the said algorithm is capable of screening out most"suspicious"medical prescriptions.