Case Studies

Our client is a startup in the intelligent document processing field, looking to create an ecosystem for insurance agencies to digitize their claim processing efforts. While the insurance claim handling is extensive and provides rich document handling functionality, our client wanted to stand out from other solutions on the market by utilizing the power of AI to automate data extraction.

Our client approached us to develop an AI data extraction module capable of automatically detecting insurance claim type and layout, extracting typed and handwritten text.

The main challenges we faced were a high variety of claim templates and designs as the layout and design of an insurance claim can look different depending on the claim type (medical insurance, business insurance, life insurance, etc.) and type of client (individuals, businesses, corporations), as well as extraction of data from complex tables.

Another important requirement for the AI module is high processing speed and high efficiency: the claim processing system is a SaaS solution for large insurance companies, meaning high document count and thousands of users daily.

We have created an AI-powered data extraction module for insurance claims that detects document structure and extracts relevant data in a matter of minutes. The module extracts data from the claims and prepares it for further processing.

We have designed the system to work efficiently under high load conditions to accommodate for the increase of users when our client turns it into a SaaS solution for other companies. We used REST API to integrate the module into an existing document processing system, facilitating easy and seamless implementation into the workflow.

Detection Of Claim Type

The module detects claim type (medical, business, etc.) and its layout to improve data extraction and its accuracy. It extracts claim type and title, automating the document identification and digitization efforts. The module also supports batch processing, meaning it can process multiple claims at once.

Processing Of Multiple Template Types

The AI module can work with various claim templates and designs, allowing for seamless data extraction from different claim types. The system can detect input fields and determine their type, like free-form text boxes, time and date fields, and more, as well as extract the data from the fields.

The system can work with claims of different designs, including complex formats that span across multiple pages, multiple claim forms on one page, input fields of irregular size and form. It not only extracts the input data, but also data type (text, date, number) and the name of the input field along with its type.

Input Field Detection

Insurance claims have many different data field types, all of which look different from each other and present relevant data differently, for example, in a form of plain text, a checkbox, date and time, etc. The AI module can detect multiple input field types, and extract data accordingly, for instance, it can detect checkboxes to determine which of the multiple options was marked.

Detection Of Tables And Table Cells

Claims our client works with often include tables with important data that needs to be extracted and processed. The module we have developed can analyze tables of any complexity, detect table cells, extract table titles and its content into an Excel file for further editing.

The AI document processing module we've developed has enhanced the functionality of the document processing platform, gaining a significant competitive advantage. The system as a whole is now used by insurance companies across the US, helping them process claims faster.