Analysis And Research On Abnormal Deep Learning Algorithm Of Medical Insurance Fraud Data

The medical insurance services is to provide citizens with reliable medical security,a social welfare system that solves the difficulties of medical treatment for the general public.In recent years,various types of medical insurance frauds and frauds have caused a huge impact on the healthy development of China’s medical insurance funds,which has seriously affected the healthy development of China’s medical system.In view of the endless emergence of medical insurance frauds,how to improve the monitoring of medical insurance frauds is the development focus of the current medical insurance system information construction.The main research contents of this article are as follows:(1)Research on evaluation indexes and identification methods of medical insurance fraud.Combined with domestic and foreign research on the detection of medical insurance fraud,there are many reasons for judging whether patients have medical insurance fraud at home and abroad,and no unified plan has been formed so far.This article selects four indexes,including the frequency of drug purchase,the total amount of all orders,the number of drug purchases and the types of drug purchases in the patient’s cost breakdown to evaluate whether the patients have committed medical insurance fraud.(2)Research on deep learning detection algorithm of medical insurance fraud.Medical insurance fraud is essentially a multi-feature classification problem,which is divided into multiple categories according to the characteristics of the data.Traditional data classification methods include clustering classification,support vector machines,decision trees,etc.The deep learning classification method used in this paper implements supervised data classification by applying deep learning,which can improve the accuracy of data classification results.(3)Research on the algorithm of medical learning fraud analysis model of medical insurance fraud.In this paper,clustering analysis algorithm and deep neural network learning model are combined to realize the classification and prediction of medical fraud.First of all,through clustering algorithm to preprocess the patient information in the patient cost details according to the evaluation indicators,and then use the labeled data is used as the input data of the DNN algorithm prediction model for training,to get a depth prediction model.Through this model,the new input data of users can be predicted,and the shortcomings of manual data annotation and manual classification can be improved.The innovations of this article are as follows:(1)A prediction model of medical insurance fraud based on deep learning is proposed.In this paper,a cluster analysis algorithm and a deep neural network learning model are used to predict patients’ medical insurance fraud.The clustering algorithm can be used to classify data,and the deep neural network model is used to build a prediction model to improve the existence of traditional prediction classification methods.Insufficient,and the model has good scalability.(2)Use Softmax logistic regression to deal with more than two label classification problems of deep learning models,that is,multi-feature multi-classification problems.In the identification of medical insurance fraud,it is necessary to combine multiple features to analyze the user’s behavior.This article will combine four indicators such as the number of patients’ medical treatment,the total amount of the total order,the number of medicines ordered and the type of medicine to determine whether the patient’s medical behavior For normal behavior,suspected fraud and fraud,this paper uses Softmax logistic regression to tune the deep learning model so that it can handle more than two label classification problems to meet the deep learning algorithm model proposed in this paper to deal with multiple features The requirements of multi-classification problems,while improving the accuracy and reliability of model prediction.In summary,this paper proposes a deep learning-based medical insurance fraud behavior prediction model,which uses a combination of clustering analysis methods and deep neural networks to accurately solve the problem of feature classification in data with high accuracy.