> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.emtelligent.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.emtelligent.com/_mcp/server.

# Introduction

This page provides a systems overview of the Extract API. This API enables customers to submit medical PDFs for processing with selected pipelines and collect the output files. The Extract API serves as an interface that triggers processing pipelines built using proprietary AI-based document solutions:

* Document Manager
* emtelliPro Medical Language Engine
* Clinical Extraction Libraries

These capabilities are orchestrated behind the scenes with optimized configurations to extract ready-to-use clinical information from medical documents.

![High-level systems overview of clinical extraction pipeline](/_fern-img/afcebfa2fad5aef670df94adf06274479cf90244d42be3c124a5d1551751d5d6.webp)

#### Links to Tutorials

* [Getting Started Tutorial](/extract/workflow/submit-and-monitor)
* [Preparing Files for Processing](/extract/workflow/preparing-files)

## Extract API Overview

At a high-level our data extraction pipelines consist of these key stages:

1. Ingestion
2. Document Splitting & Classification & OCR
3. Medical NLP Processing
4. Clinical Extractions

> **Note**
>
> Our extraction pipelines are designed to process PHI-containing real-world medical documents. All systems are run in restricted environments that are HIPAA compliant and HITRUST certified.

**Stage 1: Ingestion**

In this first stage, we:

* perform validation of the file manifest that is provided by the customer
* check that all source files listed in the manifest can be accessed before beginning the transfer
* transfer documents from AWS S3 buckets to the processing environment
* ensure that all files were transferred successfully before beginning the processing steps

**Stage 2: Document Splitting & Classification & OCR**

In this stage, the Document manager:

* splits PDF bundles into individual clinical documents
* classifies document types (e.g., "Cardiology - ECG Report") and extracts document metadata
* performs OCR text extraction

**Stage 3: Medical NLP Processing**

In this stage, the emtelliPro engine:

* processes OCR text output
* identifies clinical entities, relations and assertions

**Stage 4: Clinical Extraction**

Finally, in this stage, using the combined outputs of preceeding stages, we:

* perform complex data extraction steps for selected clinical domains such as problems, procedures, family history using Physician designed, reviewed and curated extraction libraries and scripts. See [Clinical Extraction Overview](/extract/extractions/overview)
* collate answers and produce outputs as CSV and JSONL files

After extraction and collation are complete, the output is available for retrieval by the customer. This data flow is accessed via the Extract API.