CMS Connectathon Report Out

July 21st, 2022


Functionality - Change Log

The main page displays app/server statistics, and a search dialog.

User can search value sets for parameters such as state of residence or clinical specialty.

After building a query, the data warehouse returns clinical practitioners, organizations, locations, health services, and other matched data.

FHIR query builder functional provides user explanation of which URLs are being used.

Federated Directory

By using a different settings file, the reference implementation can be run as a federated directory.  In this configuration, the server is pointed to an upstream server to hydrate and initiate its database from.  Afterwards, the federated directory can use the Subscription API to subscribe to changes.


We ran the National Directory through various security test frameworks, including Inferno and the UDAPTestTool.  The new version of Inferno identified TLS errors, which we are working with the hosting provider to resolve.  

UDAP security infrastructure remained 100% green for both Dynamic Client Registration and B2B client authentication.

Quality Control & Compliance

We ran the National Directory through the Touchstone compliance tool, to verify API compliance was maintained through the release cycle.

Data Warehouse

The data warehouse runs on a Mongo replica-set, a NoSQL storage system that is capable of scaling up to petabytes of data via clustering and sharding.  Indexes for high traffic search queries have been configured as below:

Data Quality

NPI data has been refreshed, with substantial cleanup, cross-linking, and mapping.  Practitioners (5.8m records), Organizations (1.5m records), and Locations (5.5m records) all filling search results fully and consistently.  

Hydrated the HealthcareServices collection by using the Federally Qualified Health Center (FQHC) dataset, from which we can infer basic specialties of any FQHC, such as operating a hospital, transportation via ambulances, and dental and mental health services.


The reference implementation runs on Galaxy, a Docker-as-a-Service hosting provider that runs on Amazon Web Services (AWS).  We have been running the system on a high-availability cluster (4x 2GB).

  • No labels