What challenges did Myriad Genetics face before AI implementation?

Myriad Genetics needed to reduce high document processing costs, automate key information extraction, and accelerate processing of complex medical documents. Manual extraction was a bottleneck causing delay and high labor costs. They required a scalable, accurate generative AI based document processing solution that could be integrated into existing workflows.

Source

ProductionEvidence: Medium50/100

Myriad Genetics Transforms Healthcare Document Processing with AWS Generative AI

Use case typeMedical document automationUpdated Jun 13, 2026

Myriad Genetics faced costly, slow manual document processing bottlenecks handling complex medical documents across oncology, women's health, and mental health divisions. They partnered with AWS GenAI Innovation Center to implement the open-source AWS GenAI Intelligent Document Processing (IDP) Accelerator using Amazon Bedrock foundation models, Amazon Textract, and specialized LLMs. The solution employed custom prompt engineering, model selection, and scalable serverless architecture to optimize document classification and automated key information extraction. Document classification accuracy improved from 94% to 98%, processing cost reduced by 77%, and processing time cut by 80%. Automated key information extraction achieved 90% accuracy matching human baseline, saving 78 daily labor hours and projecting $132K annual savings. The system uses Amazon Nova Pro for classification and Amazon Nova Premier for complex extraction tasks, with multimodal prompts and few-shot learning for visual context understanding.

Organization: Myriad Genetics
Industry: Healthcare
Location: United States
Published: November 2025

Reported outcomes

−77%

costCost savings

94-98%accuracy−80%time90%accuracy

Strategic outcomes

New product / capabilityAutomated key information extractionSpeed & agilityAccelerated document processing workflowsCustomer experience & trustImproved patient care through faster handlingBetter decisions & insightMore accurate document classification

Catalog median for cost savings deployments: −40% across 177 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Accuracy: 94-98% increase

AWS Machine Learning BlogNov 26, 2025Blog postInferred claimMedium evidence strength

Increased document classification accuracy from 94% to 98%.

Normalized claim

Cost: 77% decrease

AWS Machine Learning BlogNov 26, 2025Blog postInferred claimMedium evidence strength

Reduced document classification costs by 77%.

Normalized claim

Time: 80% decrease

AWS Machine Learning BlogNov 26, 2025Blog postInferred claimMedium evidence strength

Cut document classification processing time by 80%.

Normalized claim

Accuracy: 90%

AWS Machine Learning BlogNov 26, 2025Blog postInferred claimMedium evidence strength

Achieved 90% accuracy on automated key information extraction matching human baseline.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Myriad Genetics
Provider: AWS
Maturity: Production
Linked source: AWS Machine Learning Blog

Deployed scalable serverless solution integrated with existing event-driven workflows to accelerate document processing

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Document Classification
2Key Information Extraction
3Generative AI

Myriad Genetics needed to reduce high document processing costs, automate key information extraction, and accelerate processing of complex medical documents.
Manual extraction was a bottleneck causing delay and high labor costs.
They required a scalable, accurate generative AI based document processing solution that could be integrated into existing workflows.

Implemented AWS open-source GenAI IDP Accelerator, leveraging Amazon Bedrock foundation models (Nova Pro and Nova Premier), Amazon Textract, and custom prompts.
Performed prompt engineering and model selection to optimize classification and extraction accuracy and cost.
Used a multimodal approach combining document images and extracted text to enhance extraction of checkboxes and contextual information.
Deployed scalable serverless solution integrated with existing event-driven workflows to accelerate document processing.
Utilized built-in evaluation to iteratively improve the solution accuracy through configuration adjustments.

Technologies

Amazon Bedrock Amazon Nova Pro Amazon Nova Premier Amazon Textract AWS GenAI Intelligent Document Processing Accelerator

Increased document classification accuracy from 94% to 98%.
Reduced document classification costs by 77%.
Cut document classification processing time by 80%.
Achieved 90% accuracy on automated key information extraction matching human baseline.
Saved 78 daily labor hours and projected yearly savings of $132,000.
Streamlined prior authorization workflows and improved patient care through faster document handling.

Architecture

The architecture employs AWS open-source GenAI Intelligent Document Processing Accelerator built on Amazon Bedrock foundation models Amazon Nova Pro and Nova Premier, combined with Amazon Textract for OCR and multimodal prompts. It uses scalable serverless infrastructure and is integrated into existing event-driven workflows.

Sources & evidence1

Evidence: Medium50/100Evidence strength