Key Concepts

The Extract API provides a programmatic interface you will use to start jobs and get information about jobs (their status etc.). It allows customers to integrate our extraction pipelines into a their data platforms.

Key Features:

RESTful API endpoints for job submission and status tracking
Provides authenticated access
Allows asynchronous processing - submit jobs and retrieve results when ready
Performs job lifecycle management from submission to completion

Pipelines

Pipelines orchestrate the necessary processing steps. They:

Are pre-configured extraction workflows.
Can have pre-defined parameters that allow you to selectively enable or modify the pipeline behaviour. For example, the Clinical extraction pipeline supports data extraction for several core clinical domains such as problems, family history and others.

See Clinical Extractions for more information.

Jobs

A job represents a single request to process a batch of documents using a specified pipeline. Each job encapsulates all the necessary information that is needed by the pipeline. This includes:

Credentials, and
Payload information, listing documents to be processed and their metadata as a file manifest

Jobs progress through a defined lifecycle (see Job Status & Lifecycle for more details):

queued → validating → ingesting → running → completed → transferring → transfer_completed

You can poll the job status to track progress.
When complete, outputs are delivered to your destination bucket upon request.

See Output Reference for a detailed description.

Storage

The Extract API uses AWS S3 Buckets for storage. It reads and writes files to these customer-provisioned S3 buckets:

Input bucket: an S3 bucket where the you stage the documents to be processed.
Output bucket: an S3 bucket where we deliver extraction outputs to.