Automating Document Redaction for EVERSANA

Revolutionizing the WAY with Generative AI in Managing Documents

Background

Client

EVERSANA is a leader in life sciences commercialization

Challenge

To automate document redaction processes so that Health Canada PRCI could realize regulatory compliance as well as improve the efficient running of its businesses.

Solution

Introduce generative AI technologies, such as Large Language Models (LLMs), to automatically redact sensitive information from documents.

Outcome

Scalable, secure, and compliant solutions that streamlined document management and ensured audit-ready operations.

The Challenge

EVERSENANA had to automate redactions of PII and CBI on a regulatory document with an overarching requirement of high-quality workflow efficiency while being highly secure. Its manual processes involved the likelihood of human errors, and lack of transparency, and were time-consuming.

The Solution

Generative AI

Fine-tuning a foundational Large Language Model, which accurately finds and redacts PII and CBI

Reinforcement Learning

Human Feedback (RLHF) provides a feedback mechanism for continuous improvements and enhanced precision

Automating Workflows

Using AWS Lambda Serverless automation enables the redaction workflow based on the uploading of documents to an Amazon S3 bucket

High-Granular Reporting System

Ensures generation of detailed reports including the reasons behind redactions.

Secure Infrastructure

Used Amazon Cognito, API Gateway, and DynamoDB to ensure secure user authentication, access to APIs, and scalable data storage.

Streamlined Search

Key Outcomes

Efficiency Gains

Automated the redaction process, saving time and manual intervention.

Compliance

Ensured compliance with Health Canada PRCI regulations using robust NLP algorithms.

Transparency

Provided detailed reports that articulated the reasoning behind redaction decisions.

Scalability

Designed an extensible architecture that can be used to support the integration of other systems and emerging technologies.

User Satisfaction

Optimized workflows using user-friendly interfaces and API-driven processes.

Use Cases

Reporting with LLMs to generate explanations for redaction decisions
Automated extraction and export of redacted text
Classification of text as PHI, PII, and CBI
Health Canada PRCI compliance for redaction of PII and CBI
Health record processing, clinical trial documents, and research publications

Key Takeaways

AI-Powered Redaction: Utilized generative AI to revolutionize the management of documents, ensuring compliance and efficiency.
Scalable and Secure: Designed a strong architecture with AWS services that cater to the present needs and can scale for the future.
Future-Ready Framework: Included API-driven workflows and RLHF to be adaptive and continuously improve.

Technologies Used

Learn how IStream Solutions can revolutionize your operations with AI/ML technologies.

IStream Solutions collaborates with trusted partners and serves diverse clients with exceptional IT solutions. Together, we lead transformation, adaptability, and long-term impact across industries.

Building Lasting Value, Accelerating Digital Transformation Across Industries.

Quick Links

Capabilities

Reach Us

Get the latest news & updates

Technolgy & IT Solution Template Kit by Baliniz

Quick Links

Capabilities

Reach Us

Technolgy & IT Solution Template Kit by Baliniz