Accenture Creates Regulatory Document Creation Solution Using AWS Generative AI Services

This post is co-written with Accenture's Ilan Geller, Shuyu Yang and Richa Gupta.

Bringing innovative new drugs to market is a long and rigorous process. Companies face complex regulations and extensive approval requirements from governing bodies such as the US Food and Drug Administration (FDA). A key part of the filing process is the creation of regulatory documents such as the Common technical document (CTD), a comprehensive standard format document for submitting applications, amendments, supplements, and reports to the FDA. This document contains over 100 highly detailed technical reports created during the drug research and testing process. Manual creation of CTDs is incredibly labor intensive, requiring up to 100,000 hours per year for a typical large pharmaceutical company. The tedious process of collecting hundreds of documents is also prone to errors.

accent created a regulatory document creation solution using automated generative ai that allows researchers and evaluators to efficiently produce CTDs. When extracting key data from test reports, the system uses Amazon SageMaker JumpStart and other AWS ai services to generate CTDs in the appropriate format. This revolutionary approach reduces the time and effort spent creating CTDs. Users can quickly review and adjust computer-generated reports before submitting them.

Due to the sensitive nature of the data and the effort involved, pharmaceutical companies need a higher level of control, security and auditability. This solution is based on AWS well-architected principles and guidelines to enable governance, security, and auditability requirements. The easy-to-use system also employs encryption for security.

By leveraging AWS generative ai, Accenture aims to transform the efficiency of regulated industries such as pharmaceuticals. Automating the frustrating CTD documentation process accelerates the approval of new products so innovative treatments can reach patients faster. ai represents a great leap forward.

This post provides an overview of an end-to-end generative ai solution developed by Accenture for creating regulatory documents using SageMaker JumpStart and other AWS services.

Solution Overview

Accenture created an ai-powered solution that automatically generates a CTD document in the required format, along with the flexibility for users to review and edit the generated content. The preliminary value is estimated at a 40% to 45% reduction in creation time.

This ai-based generative solution extracts information from technical reports prepared as part of the testing process and delivers the detailed dossier in a common format required by central governing bodies. Users then review and edit the documents, where necessary, and submit them to central governing bodies. This solution uses the SageMaker JumpStart AI21 Jurassic Jumbo Instruct and AI21 Summarize models to extract and create the documents.

The following diagram illustrates the architecture of the solution.

The workflow consists of the following steps:

1. A user accesses the regulatory document creation tool from their computer's browser.
2. A React application is hosted on AWS Amplify and accessed from the user's computer (for DNS, use Amazon Route 53).
3. The React app uses the Amplify authentication library to detect if the user is authenticated.
4. Amazon Cognito provides a local user pool or can be federated with the user's active directory.
5. The application uses Amplify libraries for Amazon Simple Storage Service (Amazon S3) and uploads user-provided documents to Amazon S3.
6. The application writes the job details (application-generated job ID and Amazon S3 source file location) to an Amazon Simple Queue Service (Amazon SQS) queue. Captures the ID of the message returned by Amazon SQS. Amazon SQS enables a decoupled, fault-tolerant architecture. Even if some backend errors occur when processing a job, having a job log within Amazon SQS will ensure that retries are successful.
7. Using the job ID and message ID returned by the previous request, the client connects to the WebSocket API and sends the job ID and message ID to the WebSocket connection.
8. WebSocket triggers an AWS Lambda function, which creates a record in Amazon DynamoDB. The record is a key-value mapping of the job ID (WebSocket) to the connection ID and message ID.
9. Another Lambda function is triggered by a new message in the SQS queue. The Lambda function reads the job ID and invokes an AWS Step Functions workflow to process data files.
10. The Step Functions state machine invokes a Lambda function to process the source documents. The function code invokes Amazon Textract to parse the documents. Response data is stored in DynamoDB. Depending on specific data processing requirements, data can also be stored in Amazon S3 or Amazon DocumentDB (with MongoDB support).
11. A Lambda function invokes the Amazon Textract DetectDocument API to parse tabular data from source documents and stores the extracted data in DynamoDB.
12. A Lambda function processes data based on mapping rules stored in a DynamoDB table.
13. A Lambda function invokes message libraries and a series of actions using generative ai with a large language model hosted through Amazon SageMaker for data summarization.
14. The document writer Lambda function writes a consolidated document to an S3 processed folder.
15. The job callback Lambda function retrieves the callback connection details from the DynamoDB table and passes the job ID. The Lambda function then makes a callback to the WebSocket endpoint and provides the link of the processed document from Amazon S3.
16. A Lambda function removes the message from the SQS queue so that it is not processed again.
17. A document generator web module converts JSON data into a Microsoft Word document, saves it, and renders the processed document in the web browser.
18. User can view, edit and save the documents to S3 bucket from the web module. This helps in necessary revisions and corrections, if any.

The solution also uses SageMaker notebooks (labeled T in the architecture above) to perform domain adaptation, tune models, and deploy SageMaker endpoints.

Conclusion

In this post, we show how Accenture uses AWS Generative ai services to implement an end-to-end approach to a regulatory document creation solution. In early testing, this solution demonstrated a 60% to 65% reduction in the time required to create CTDs. We identified gaps in traditional regulatory governance platforms and augmented generative intelligence within their framework for faster response times, and are continually improving the system as we engage with users around the world. Contact the Accenture Center of Excellence team to dive deeper into the solution and implement it for your clients.

This joint program focused on generative ai will help increase time to value for joint Accenture and AWS customers. The effort builds on the companies' 15-year strategic relationship and uses the same proven levers and accelerators created by the Accenture AWS Business Group (AABG).

Connect with the AABG team at [email protected] to drive business results by transforming to an intelligent data enterprise on AWS.

For more information about Generative ai on AWS using Amazon Bedrock or SageMaker, see Generative ai on AWS: technology and Getting Started with Generative ai on AWS using Amazon SageMaker JumpStart.

You also can ai-interest-learn.html” target=”_blank” rel=”noopener”>Subscribe to the AWS Generative ai Newsletterincluding educational resources, blogs and service updates.

About the authors

Ilan Geller He is a managing director in the data and artificial intelligence practice at Accenture. He is the AWS Global Partner Leader for Data and ai and the Center for Advanced ai. His roles at Accenture have primarily focused on the design, development and delivery of complex data, ai/ML and, more recently, generative ai solutions.

Shuyu Yang He is a Large Language Model and Generative ai Delivery Leader and also leads the Accenture ai (AWS DevOps Professional) CoE (Center of Excellence) teams.

Richa Gupta He is a technology architect at Accenture and leads several ai projects. He has over 18 years of experience architecting scalable ai and GenAI solutions. His area of ​​expertise is ai architecture, cloud solutions and generative ai. He plays a pivotal role in various pre-sales activities.

Shikhar Kwatra is a solutions architect specializing in ai/ML at Amazon Web Services and works with a leading global systems integrator. He has earned the title of one of the youngest Indian master inventors with over 500 patents in the domains of ai/ML and IoT. Shikhar assists in architecting, building and maintaining scalable and cost-effective cloud environments for the organization, and supports GSI partner in building strategic industrial solutions on AWS. Shikhar enjoys playing the guitar, composing music, and practicing mindfulness in his spare time.

Sachin Thakkar is a Senior Solutions Architect at Amazon Web Services and works with a leading Global Systems Integrator (GSI). He brings more than 23 years of experience as an IT Architect and technology Consultant for large institutions. His focus area is data, analytics and generative ai. Sachin provides architectural guidance and supports GSI partner in building strategic industrial solutions on AWS.