Guardoc Health uses Amazon Nova to process medical documents
Amazon Web Services
Guardoc Health uses Amazon Nova models on Amazon Bedrock to transform clinical documentation in long-term care. The system reduces documentation errors by 46%, audit fines by 70%, and delivers over $400K annual ROI per facility.
Guardoc Health leverages the Amazon Nova family of models, available through Amazon Bedrock, to process medical documents for skilled nursing facilities and assisted living centers. The pipeline uses Retrieval Augmented Generation (RAG) for medical condition classification, combining Amazon Textract for OCR, Amazon Titan Text Embeddings for vector embeddings, and Amazon Nova Pro for multimodal reasoning over raw PDFs. For PDF processing, Amazon Nova Pro handles layout interpretation, checkbox states, signatures, and handwriting. A hybrid OCR pipeline pairs Amazon Textract with Amazon Nova models for medication extraction, tackling structured tables, handwritten additions, and complex formats. Guardoc reports a 46% reduction in documentation errors, 70% fewer audit fines, and over $400K annual ROI per facility. On peak days, the system processes over 1 million documents, with traceability to specific source pages for clinical safety.
- Сокращения
- RAG = Retrieval Augmented Generation — генерация с дополнением извлечением
- OCR = Optical Character Recognition — оптическое распознавание символов
- ROI = Return on Investment — возврат инвестиций
- k-NN = k-nearest neighbor — k ближайших соседей
- PDPM = Patient-Driven Payment Model — модель оплаты, ориентированная на пациента
Source: AWS ML blog —
original