Review insurance claim data and identify potential fraud scenarios with Amazon Q for Business
The goal of this post is to provide a GenAI tool to accelerate Insurance Claims Specialist productivity in reviewing claims incident data from an insurance settlement demand letter and provide suggestions for potential fraud scenarios
Surya Vangala
Amazon Employee
Published Sep 18, 2024
Last Modified Sep 24, 2024
A settlement demand letter is a formal request for compensation that is sent by an individual (the claimant) or their attorney to the party responsible (the insured) for an injury or their insurance company. It is often used in personal injury cases (from Automotive or other incidents) to help victims get compensation faster than a legal settlement.
A settlement demand letter typically includes:
- Incident details: A description of the accident, including the date, time, and location
- Injuries: A description of the injuries and their effects on the victim's daily life
- Treatment: A timeline of treatment, including the type of treatment and recovery progress
- Medical bills: A list of medical bills and lost income statements
- Liability: An explanation of why the policyholder/insured is liable for the victim's injuries
- Compensation: The amount of compensation demanded from the insurance company
However, there is no fixed format for these letters and are often unstructured with claim details mentioned across different sections of the document. Insurance claims specialists spend significant amount of time manually reviewing the settlement demand letter documents and subsequently perform background verification for each of the claims from third party sources. Also too often these letters have fraudulent, fabricated or exaggerated information about the incident and the declared damages to individuals and property. The proposed chat solution described in this post built with Amazon Q Business helps accelerate the review of these documents by quickly summarizing and extracting relevant data fields from the document while also suggesting possible fraud scenarios guiding the claims specialist to validate the verity of the claims.
Amazon Q Business is a generative AI–powered assistant that can answer questions, provide summaries, generate content, and securely complete tasks based on data and information in your enterprise systems. It empowers employees to be more creative, data-driven, efficient, prepared, and productive.
High Level Architecture of the solution with Amazon Q Business
This solution uses AWS IAM Identity Center to grant access to users to the Amazon Q application. Create a user for this application in IAM Identity Center by referring to the steps here.
Navigate the AWS Console and search for and select the Amazon Q Business service from the console top search bar
Select "Create application" and provide an "Application name", continue with the default selections and select "Create"
Continue with "Use native retriever" option selected for Retrievers and select "Starter" for Index provisioning for this demo workload (select Enterprise if building for Production workloads). Leave the number of units as 1 and select "Next"
On the Connect data sources screen, do not select any source and select "Next"
On the Manage access screen, select Users tab and select "Add groups and users". Select "Assign existing users and groups" option and select "Next" and "Get started". Search for the user already created in the IAM Identity Center as part of the prerequisite user setup above and select "Assign"
Select "Done" to complete the Amazon Q Business application creation
Select the Web Experience URL to launch the Amazon Q Business application login screen. Enter the username of the IAM Identity Center user selected above and select "Next". Enter the password and select "Sign in"
Amazon Q Business application is displayed with chat interface.
Select to upload files and upload a sample Settlement Demand Letter (available on github). Review the sample letter below.
Query the Amazon Q chat interface to "Summarize the demand letter" that was uploaded. The results display a summary of the demand letter PDF document which shows key sections of the document
List potential fraud scenarios in the document using a chat question like: "Identify possible fraud scenarios given the information in the document"
Explore key data fields in the document further using the suggested sample chat questions:
- What was the date of loss ?
- What was the location of the incident ?
- Who is the claimant ?
- Who is the insured ?
- What is the total demand amount ?
As demonstrated in this post, Amazon Q Business allows insurance claims specialists, developers, IT engineers, and business teams with minimal to no coding skills to quickly create an enterprise grade GenAI based chat application to improve their productivity in minutes. This application can be used across Financial Services, Healthcare, Media, and numerous other industries where documents need to be manually reviewed and processed for claims and other use cases.
Any opinions in this post are those of the individual author and may not reflect the opinions of AWS.