We developed an end-to-end claims processing digitization solution to reduce manual efforts and empower predominant health insurer with Gen AI driven automationsdata

CHALLENGE

Customer is facing critical challenges in their claim processing workflow, hindering efficiency and accuracy. Key obstacles include:

• The current manual process, involving document segregation and data entry, is time-consuming and prone to errors.
• The wide array of inconsistent document formats from various healthcare providers complicates data extraction and interpretation.
• The absence of an in-house OCR tool limits the ability to quickly and accurately digitize crucial information from physical documents

They want to implement an intelligent, automated system capable of handling diverse document formats, extracting relevant data with high accuracy, and employing advanced analytics for smarter claim processing.

Why were we brought in?

Ganit has made a significant dent in various industries using data science and analysis. Ganit partners with clients to translate their data into a tangible, insightful plan of action that delivers on a measurable impact to the clients’ topline & bottom-line growth.

With a strong Gen AI practice and a dedicated Gen AI Centre of Excellence (CoE), Ganit has advanced capabilities in building consumable Gen AI driven solutions for our customers. With almost 20 demos catering to different industry verticals and business functions, Ganit has a proven track of delivering customer delight through accelerated implementation.

Solution

Using the power of machine learning and cloud computing, a scalable AI-powered automated claims processing system was developed, leveraging AWS services to revolutionize their workflow.

• An incoming document in zip format is placed in an S3 bucket, which triggers a preprocessing Lambda function
• This Lambda function extracts files from the zip, which on average contains 15-20 documents, converts each file to an image, and stores them in a designated S3 bucket
• Each stored image triggers a new Lambda function that performs OCR, extracting text and bounding boxes
• The extracted text and bounding boxes are sent via a SageMaker trigger Lambda to a custom segregation model deployed on a SageMaker endpoint with over 96% accuracy; classification predictions and metadata are saved in a database
• Users verify these predictions in an application developed by Ganit, and a QC verification button initiates a post-processing Lambda
• This post-processing Lambda consolidates images of the same class into a single PDF and stores it in S3
• The consolidated PDF triggers another Lambda that calls Claude Haiku and Mistral 7B LLM models via Bedrock API; the post-processed results are displayed as key-value pairs in the application, allowing users to make updates, which are then pushed to an internal application

Features of the tool

• The AI-powered system automates document handling, data extraction, and classification, drastically reducing processing times, minimizing human errors and ultimately improving customer satisfaction
• Solution ensures enhanced accuracy in data extraction and interpretation across diverse document formats while maintaining regulatory compliance
• Solution offers unparalleled scalability to handle growing data volumes and evolving document types