Scout#

Welcome to the Scout documentation!

Scout is a data exploration and clinical insights platform designed to help users access and analyze large volumes of medical imaging data. Data are processed and ingested into a data lake where they are made available for exploration and analysis through multiple interfaces.

The first release of Scout is focused on HL7 radiology reports. Future versions will incorporate DICOM, pathology reports, and extracted features.

Quickstart#

Scout Launchpad#

After authentication, your initial landing page is the Scout Launchpad. From here, you can access all Scout services, including Analytics, Notebooks, Chat (if enabled), and this documentation site.

Scout Launchpad

Core Services#

Scout provides three primary interfaces for working with data:

Scout Analytics#

Selecting Analytics from the Launchpad takes you to Apache Superset, a powerful data visualization and exploration tool. Your initial landing page is the Scout Dashboard, designed to give you an overview of all report data. From here, you can analyze and explore the data either graphically using the visualization builder or via SQL Lab for direct SQL queries. See the Trino SQL documentation for SQL syntax reference.

Scout Dashboard

Scout Chat#

Selecting Chat from the Launchpad provides an AI-powered interface for natural language querying of radiology reports. Ask questions in plain English and receive data-driven answers powered by large language models with direct access to the Scout Delta Lake. Learn more in the Chat documentation.

Note: The Chat service is optional and may not be available in all Scout deployments.

Scout Chat

Scout Notebooks#

Selecting Notebooks from the Launchpad launches JupyterHub with notebooks that query the data lake through Trino for fully customizable data analysis. An example Jupyter notebook with sample code to access and analyze data is provided in Scout/Quickstart.ipynb. This notebook provides example code using pandas with SQL to search for reports, filter by various criteria, and export results to CSV files.

Scout Quickstart Notebook

Next Steps#

  • Data Schema: Understand the structure of report data in the data lake and the mapping of HL7 fields to table columns

  • Services: Learn about the services that make up Scout, including user-facing tools (Analytics, Notebooks, Chat) and backend infrastructure (Temporal, Delta Lake, MinIO, monitoring)

  • Chat Guide: Explore the AI-powered chat interface for natural language queries (if enabled in your deployment)

  • Authentication: Learn about Scout’s single sign-on authentication and user approval process

  • Data Authorization: Per-user row filtering and column masking — how Scout decides what data each user can see, and how administrators configure it

  • Report Ingestion: Understand how HL7 reports are ingested and processed into the data lake

  • Tips & Tricks: Discover helpful hints for using Scout services effectively

  • Technical Documentation: Deep dive into technical aspects of the Scout system

  • Report Issues: Submit issues via Jira