Use an AI model to predict the authenticity of an insurance claim

Use an AI model to predict the authenticity of an insurance claim

Requirement

Build an AI model to predict if an insurance claim is genuine.

Use case

An insurance application contains a form — Fraud Detection. To avoid being inundated by false insurance claims, whenever an entry is made to the form, an AI model automatically predicts if the submitted claim is fraudulent.
​
See how it works

Steps to follow

1. Create a form with the following details:
Form
Form Link Name
Field Type
Field Name
Field Link Name
Fraud Detection
Fraud_Detection
Single Line
Incident Severity
Incident_Severity
Currency
Total Claim Amount
Total_Claim_Amount
 
 
Dropdown
  • None
  • Police
  • Ambulance
  • Fire
  • Other
 
Authorities Contacted
Authorities Contacted
 
 
Number
Months As Customer
Months_As_Customer
 
 
Radio
  • Yes
  • No
Is Fraud Claim?
Is_Fraud_Claim

2. Add a prediction field with the field name and field link name Likely to be Fraud and Likely_to_be_Fraud, respectively, to the Fraud Detection form with the following details:

3. Create a workflow with the following details:

Click Add New Action, and add the following snippet:
  1. disable Likely_to_be_Fraud;

4. Import training data to the Fraud Detection form. This helps the prediction field to analyze the data in the form and use machine learning to predict a field value for all future records.
 
Sample training data is attached as a file to this article.

5. Navigate to AI models under Microservices and retrain the Likely to be Fraud model. By doing so, the AI model will be trained by considering the training data that we just imported.

See how it works