TWM549931U - Insurance policy risk assessment system - Google Patents

Description

Policy Claims Risk Prediction System

The present invention relates to a risk prediction system, and more particularly to a risk prediction system for policy claims.

With the improvement of the knowledge level, people's purchase of insurance has become more and more popular, and the insurance company's insurance sales performance has also increased. Although the insurance sales performance has gradually increased, the number of applications for insurance claims has increased relatively, and the burden on insurance companies in the inspection of insurance claims has also increased relatively, especially for the inspection of insurance claims for fraud; for example, accidents against fraud, An audit of the diagnosis of a fake disease.

At present, for each claims case filed by the policyholder/policy, the insurance company relies on the experience of the insurance practitioner to judge whether the claim event has the possibility of fraud and further investigate the possible claims. Therefore, the labor cost required is quite large.

Therefore, the purpose of the present invention is to provide a policy claim risk prediction system.

Therefore, the new policy claim risk prediction system is used to predict the claim score for the policy of the policy holder, and includes a storage module and a processing module.

The storage module stores a predictive model trained by the policy data of the plurality of policies, wherein the policy information of each policy includes relevant information of the recruiter of the policy, relevant information of the delivery service personnel of the policy, and the Information about the current status of the policy.

The processing module is configured to use the prediction model to calculate the claim score of the policy according to the policy information of the policy of the policy holder.

The effect of this new model is that it can predict the claim risk of the policy.

Referring to FIG. 1 , an embodiment of the new policy claim risk prediction system is applicable to an insurer, and includes a storage module 11 , a processing module 12 , and an input and output module 13 .

The insurer's practitioners can operate the policy claims risk prediction system through the input and output module 13.

The storage module 11 stores a plurality of predictive models for claims fraud prevention, which are trained in advance according to different types of insurance (all types of insurance, medical insurance) and claims types (diseases, accidents), including Four prediction models, such as a full-risk disease claim prediction model, a full-risk accident claims prediction model, a medical insurance disease claims prediction model, and a medical insurance accident compensation prediction model; preferably, each prediction model is a regression model, A type of neural network model, or a decision tree.

Each predictive model is a predictive model trained based on policy information of a plurality of policies prepared in advance, that is, so-called training materials. The policy information of each policy includes the relevant information of the recruiter of the policy, the relevant information of the delivery service personnel of the policy, the relevant information of the current status of the policy, and the claim history data of the policyholder of the policy. The relevant information of the recruiter of the policy includes the characteristic parameters of the refusal rate, the refusal rate, the claim amount, the number of claims, etc. of the recruiter; the relevant information of the delivery service personnel of the policy includes the refusal of the delivery service personnel Characteristic parameters such as rate, refusal rate and claim amount; the relevant information of the current status of the policy includes the current validity of the policy, the amount of the insured and the attribute of the policy holder's relationship with the policy.

The claim history data of each policyholder includes relevant information about the accident of the policyholder, relevant information of the medical examination center corresponding to the accident of the policyholder, and the number of times the policyholder receives the claim, the hospitalization amount, the medical record, etc. Characteristic parameters; wherein the relevant information of the accident occurred by the policyholder includes the characteristics of the disease/surgery type, the length of hospital stay, the accident target, and the relevant information of the medical examination center corresponding to the accident of the policy holder includes the rejection of the medical examination institute Characteristic parameters such as odds, regional level and the number of operations of the policy holder in the medical examination center.

Referring to FIG. 2, when the above-mentioned pre-trained predictive models for claims fraud prevention are actually applied, the insurer's practitioners operate through the output module 13 for each policy/case of claim settlement. The policy claim risk prediction system selects a predictive model applicable to the policy from the predictive models for claim fraud based on the policy type of the claim case (step 21). Then, the insurer's practitioner further operates the policy claim risk prediction system, so that the processing module 12 extracts a plurality of feature parameters corresponding to the selected prediction model from the policy information of the policy as input of the selected prediction model. Thus, a claim risk score for a pair of policies that should be applied for claims is calculated based on the feature parameters (step 22).

In particular, the higher the claim risk score, the higher the probability that the claim case is fraudulent. If the claim risk score is greater than a predetermined threshold, the insurer further investigates whether the claim event is fraudulent. That is to say, with the new policy claims risk prediction system, the insurer can help the insurance company to remove the policy/case of the claim that has a lower probability of fraud and filter out the policy/case of the claim with higher possibility of fraud. The industry only needs to evaluate and investigate the policies/cases for claiming claims with high probability of fraud, which can greatly save the labor cost for checking whether the claims event is fraudulent.

It is worth mentioning that in addition to the use of the client's policy information to train these predictive models for claims fraud prevention, the client's policy information can also be used to train a first predictive model for fast claims and a second prediction model; wherein the first prediction model and the second prediction model are respectively used to calculate a quick claim score for a policy of a claim case of a customer of the claimless history data and the claim history data. In particular, the higher the quick claim score, the lower the probability that the claim case will be fraudulent, and the less necessary for the insurer to further investigate whether the claim event is fraudulent. If the quick claim score is greater than a predetermined threshold, the insurer does not investigate whether the claim is fraudulent and provides customer claims quickly.

In summary, the new policy claim risk prediction system pre-trains a predictive model for claims fraud prevention or for quick claims, and extracts corresponding feature parameters for a policy/case for claim settlement. And using the prediction model to calculate a claim risk score or a quick claim score corresponding to the policy/case of the claim for claim based on the feature parameters, and further predicting the possibility of fraud based on the claim risk score or the quick claim score A high or low policy/case for applying for a claim, so it is indeed possible to achieve the purpose of this new type.

However, the above is only the embodiment of the present invention, and when it is not possible to limit the scope of the present invention, all the simple equivalent changes and modifications according to the scope of the patent application and the contents of the patent specification are still This new patent covers the scope.

11‧‧‧ Storage Module
12‧‧‧Processing module
13‧‧‧Output module
21~22‧‧‧Steps

Other features and effects of the present invention will be apparent from the following description of the drawings, wherein: FIG. 1 is a block diagram illustrating an embodiment of the present policy claim risk prediction system; and FIG. 2 is a flow Figure, illustrating the operational flow of the policy claims risk forecasting system.

11‧‧‧ Storage Module

12‧‧‧Processing module

13‧‧‧Output module

Claims (8)

A policy claim risk prediction system for predicting a claim score for a policy of a policyholder, and comprising: a storage module storing a predictive model trained by policy data of multiple policies, wherein policy information of each policy Information relating to the recruiter of the policy, relevant information of the delivery service personnel of the policy and relevant information of the current status of the policy; and a processing module for utilizing the predictive model according to the policy of the policyholder The policy information calculates the claim score corresponding to the policy. The policy claim risk prediction system according to claim 1, wherein the relevant information of the recruiter of the policy includes the refusal rate, the refusal rate, the claim amount, and the number of claims of the recruiter, and the relevant service personnel of the policy are related to The information includes the refusal rate, the refusal rate and the claim amount of the delivery service personnel, and the relevant information of the current status of the policy includes the current validity of the policy, the amount of the insurance to be insured, and the attribute of the related person of the policy holder. The policy claim risk prediction system according to claim 1, wherein the policy information of each policy further includes the claim history data of the policy holder. The policy claim risk prediction system according to claim 3, wherein the claimant's claim history data includes relevant information of the accident of the policyholder, relevant information of the medical examination center corresponding to the accident of the policyholder, and the insurance The number of times the household gets the claim, the amount of hospital stay, and the medical record. The policy claim risk prediction system according to claim 4, wherein the relevant information of the accident occurred by the policyholder includes the disease/surgery type, the length of hospital stay, the accident target, and the relevant information of the medical examination center corresponding to the accident of the policy holder It includes the refusal rate of the medical examination center, the district level and the number of operations of the policy holder in the medical examination center. The policy claim risk prediction system of claim 1, wherein the predictive model is a regression model. The policy claim risk prediction system according to claim 1, wherein the predictive model is a type of neural network model. The policy claim risk prediction system of claim 1, wherein the predictive model is a decision tree model.