本科毕业设计(论文)
外文翻译
Medical Insurance Fraud Recognition Based on Improved Outlier Detection Algorithm
作者:JIAN WU, RUNTONG ZHANG, XIAOPU SHANG and FUZHI CHU
国籍:Beijing, China
出处:2017 2nd International Conference on Artificial Intelligence and Engineering Applications (AIEA 2017) ISBN: 978-1-60595-485-1
原文正文:
ABSTRACT
This paper presents an improved outlier detection algorithm based on K-means clustering to identify suspicious medical fraud in medical insurance audits. This paper
elaborates how to preprocess medical insurance data for medical insurance fraud, and
puts forward the improved principle and process of outlier detection algorithm based
on K-means clustering. The experiment is carried out by using real medical insurance
data, and the efficiency of the algorithm conducted a test.
KEYWORDS
Medical insurance, K-means clustering, Outlier algorithm.
INTRODUCTION
Medical insurance is an insurance to compensate for the medical expenses of medical treatment. Medical insurance system is to solve the insured treatment problem, the use of health insurance fund. Internationally, it is generally agreed that the average proportion of health insurance violations is between 20-30%. In recent years, medical insurance system reform to further promote the scope of basic medical insurance services is growing. With the widespread use of medical insurance and the widespread use of health insurance card, accompanied by the emergence of some health insurance card non-compliance, such as over-treatment, over-inspection, over
service and other acts, the entire medical insurance industry caused great harm. With
the popularity and maturity of medical information systems, all medical units have kept a lot of data and records, while the data is still increasing, but health insurance for the violation of the monitoring still remain in the main stage of labor. So the use of
these data efficiently and automatically detect violations will be of great significance.
The main purpose of outlier detection is to detect anomalous or abnormal data from a given data set. As an important data analysis technology in data mining, outlier detection technology has been widely used in network intrusion detection, financial
fraud detection and other fields. Therefore, this paper attempts to study the suspicious
behavior of medical insurance from the point of view of outlier detection of data mining.
After normalizing the medical insurance data, the improved K-means clustering
algorithm based on K-means clustering is used to solve the suspicious medical
insurance data in the data was identified.
LITERATURE REVIEW
Medical Insurance Fraud Detection
In foreign countries, there are many studies that will apply data mining to health
insurance fraud. IBM Research Center Marisa first proposed to use data mining method to detect health insurance fraud, and the use of association rules and neural segmentation method to identify medical insurance fraud. Taiwan Yang proposed a detection framework based on the 'clinical path', presented a pattern of 'behavioral events', found patterns by digging frequent subgraphs, and experimented with Taiwan National Health Insurance data, the results show that the model is more effective than manual inspection. In the intelligent method, the neural network is widely used to establish the fraud recognition model because of its special merit. He designed a neural network model to detect medical insurance fraud, reaching 88.4% correct rate. He also applied the genetic algorithm and KNN algorithm to health insurance fraud detection, which was used by the Australian Health Insurance Board to identify medical insurance fraud as the law further improved accuracy.
As Chinas health insurance system and foreign countries there are some differences, has its own characteristics, therefore, many domestic scholars with Chinas health insurance policy to explore the actual situation. For example, He Junhua from the three data mining three points of view, respectively, based on the
clustering method of the insured person subdivision model, based on sequence pattern
discovery mode of mining and frequent pattern mining algorithm based on consistent
fraud Behavior detection. Liu Jiangchao defined the abnormal prescription as a prescription with less than a certain threshold for the combination of commonly used
drugs. The study used frequent itemsets mining to excavate the commonly used drug
combination in medical insurance data, and obtain some meaningful drug patterns to
explain the analysis. Yuan Xiaodong proposed a clinical behavior anomaly detection model based on association rules. Based on the characteristics of the temporal behavior characteristics of clinical behavior data, a frequent sequence mining algorithm with time constraints was improved, and the association rules of clinical behavior sequence were obtained by algorithm. Based on the construction of clinical behavior abnormalities detection model.
K-means Algorithm
The clustering point-based value detection method based on clustering is that it
does not require pre-tagging data and is combined with clustering algorithms to detect
outliers. ROCK, DBSCAN and BIRCH clustering algorithms focus only on clustering
data and do not have ex
剩余内容已隐藏,支付完成后下载完整资料
英语原文共 8 页,剩余内容已隐藏,支付完成后下载完整资料
资料编号:[273074],资料为PDF文档或Word文档,PDF文档可免费转换为Word
以上是毕业论文外文翻译,课题毕业论文、任务书、文献综述、开题报告、程序设计、图纸设计等资料可联系客服协助查找。