Guardoc Health Automates Clinical Documentation with Amazon Nova AI Models

Guardoc Health processes over one million clinical documents daily using Amazon Nova models through Bedrock, representing a significant advancement in healthcare documentation automation and clinical accuracy.

The Critical Balance: AI-Powered Clinical Documentation

Implementing AI in clinical documentation requires careful risk assessment. Documentation errors can cascade into serious consequences:

• Denied Medicare claims under the Patient-Driven Payment Model (PDPM)
• Substantial audit fines and regulatory penalties
• Litigation exposure from documentation discrepancies
• Critical patient safety risks from missed conditions affecting treatment decisions

However, successful implementation delivers measurable benefits: reduced correction cycles, fewer hospital transfers, and significantly lower compliance costs. Guardoc Health's documentation software for long-term care providers demonstrates these outcomes through published deployment data.

Understanding the Documentation Challenge

Guardoc Health's processing pipeline handles documents in virtually every clinical format:

Multi-page PDFs combining handwritten physician annotations with printed text, prior authorization forms where single checkbox states determine coverage decisions, medication lists appearing as structured tables or free text, and patient intake forms mixing typed fields with rubber stamps and handwriting.

Research published in BMJ Quality and Safety estimates approximately 12 million US outpatients annually are affected by diagnostic errors, with information-handling failures identified as contributing factors. At Guardoc's processing volume, even a one percent error rate in condition detection would generate thousands of incorrect records daily, each carrying patient safety or compliance implications.

Quantified Performance Improvements

Guardoc Health reports substantial operational improvements:

In a quarterly deployment covering two facilities and 200 patients, the system achieved:

• 847 documentation corrections implemented
• 86 PDPM reimbursement accuracy issues identified and resolved
• 74% reduction in hospital transfers per 100 admissions

A separate case study encompassing seven facilities and 1,618 residents identified 10,612 documentation issues, according to company data.

Cost-Optimized AI Architecture

Guardoc's technical architecture implements Retrieval Augmented Generation (RAG) for condition classification, extracting evidence from patient documentation before applying reasoning to generate conclusions.

The processing pipeline follows a tiered approach:

1. Amazon Textract extracts text and structural metadata from incoming pages at the lowest per-page cost point
2. Content is chunked along clinical boundaries, preserving medication lists and diagnosis sections as intact units
3. Each chunk is embedded using Amazon Titan Text Embeddings V2 and stored in Amazon DynamoDB with patient-level partitioning
4. A custom pre-filter narrows candidate sets by document type and recency before k-nearest neighbor search retrieves relevant chunks
5. Amazon Nova 2 Lite performs text-based filtering to eliminate obvious non-matches
6. Amazon Nova Pro receives raw PDF bytes for final multimodal reasoning over layout, handwriting, signatures, and stamps

This design philosophy prioritizes cost efficiency by reserving computationally intensive multimodal reasoning for the final stage where it delivers maximum value, while cheaper components handle high-volume embedding and filtering operations.

Addressing Complex Documentation Scenarios

Two document categories represent the most challenging cases:

Medication extraction presents additional complexity. Drug names, dosages, administration routes, and frequencies appear in multiple formats: structured tables, prose within physician notes, handwritten additions to printed lists, and multi-generation faxed scans.

Guardoc's hybrid pipeline addresses this by running Amazon Textract first for clean printed tables, then passing both the original PDF and Textract output to Amazon Nova Pro to resolve wrapped table columns, handwritten additions, and non-standard formats that OCR alone cannot accurately parse.

Industry Impact and Future Direction

"With the Nova family, we're making it easier for healthcare organizations to detect high-risk cases earlier and act before issues become costly. By automating workflows that once required manual oversight, the Nova family helps teams reduce compliance gaps, prevent errors, and focus more of their time on improving patient outcomes."

— Assaf Amiaz, Director of Product at Guardoc Health

The deployment demonstrates how AI-powered document processing can transform clinical documentation workflows, enabling healthcare teams to shift from manual oversight to proactive patient care while maintaining compliance standards and reducing operational costs.

Key takeaways for healthcare organizations:

• Strategic AI implementation can significantly reduce documentation errors and compliance risks
• Cost-optimized architectures make large-scale document processing economically viable
• Multimodal AI models effectively handle complex clinical documentation formats
• Automated workflows enable healthcare professionals to prioritize patient care over administrative tasks