Introduction

This repository contains scripts and data for creating containers that INDICATE data providers can deploy in order to implement the evaluation of quality indicators as well as communication with the central hub for the INDICATE quality benchmarking dashboard.

Containers

This project creates the following container images:

1. The database container contains a PostgreSQL database initialized with OMOP CDM tables and a collection of terminology. At runtime, the database is used to store the results of evaluating quality indicators. The stored results can be used for two purposes:

   1. The data provider can inspect conformance to the INDICATE quality indicators on a per-patient level.
   2. The stored results are aggregated via views into a ward-level representation. This aggregated data is sent (without identifiable information) to the central benchmarking service for distribution to all data providers.

2. The cql-on-omop container contains an evaluation engine for CQL libraries. This engine applies CQL libraries which implement the INDICATE quality indicators to a database containing clinical data (which is distinct from the database in the database container). The results computed by the CQL libraries are stored in the database container for local inspection, aggregation and distribution as explained above.

3. The data-exchange container contains a simple program that uploads aggregated quality indicator results to the INDICATE central hub.. To this end, the program fetches aggregated quality indicator results from the database container and uploads them, together with a unique id for the data provider, to the hub node of the INDICATE quality indicator dashboard architecture.

4. The dashboard container contains the web-based quality indicator dashboard application that is the main purpose of the whole architecture. At the moment, the dashboard container communicates with the INDICATE central hub to retrieve aggregated quality indicator results from all data providers and present the results in different web-based views. Later, the dashboard container may be extended to also communicate with the database container in order to display patient-level quality indicator results that are specific to the data provider and must also not leave the systems of the data provider for privacy and data protection reasons.

Dependencies

• PostgreSQL

• https://github.com/umg-minai/cql-on-omop

• https://github.com/aptible/supercronic

• INDICATE CQL Libraries

Build

Obtain Dependencies

1. cql-on-omop

   This dependency will be obtained automatically from the GitHub Container Registry at https://github.com/umg-minai/cql-on-omop/pkgs/container/cql-on-omop.

2. INDICATE Quality Indicator CQL Libraries

   TODO

3. INDICATE Data Exchange Client

   This dependency will be obtained automatically from the GitHub Container Registry at https://github.com/umg-minai/indicate-data-exchange-client/pkgs/container/indicate-data-exchange-client.

4. Dashboard

   This dependency will be obtained automatically from the GitHub Container Registry at https://github.com/umg-minai/indicate-dashboard/pkgs/container/indicate-dashboard.

Set up Database Access

1. Add connection data for the OMOP database which provides the clinical data to databases.env in the root directory of this repository.

2. Enter the database password for that database into the file source-database-password in the root directory of this repository.

3. Choose an arbitrary password for the internal database in the database container and enter that password into the file target-database-password in the root directory of this repository. One way to do this is

   echo INDICATE | mkpasswd -m descrypt -s > target-database-password

Set up Data Exchange (optional)

If default settings for communication with the INDICATE data exchange server are not appropriate, customize the settings as described below.

1. TODO DATA_EXCHANGE_ENDPOINT

Set a Provider Display Name

Some views of the dashboard container display quality indicator results for multiple data provider and highlight the results for the data provider running the dashboard. This works because the data provider (and only the data provider) can recognize its "own" results based on the unique provider id. This highlighting can be augmented by specifying a human-readable name that should be displayed alongside the highlighted results. The chosen name, which is an arbitrary string, should be written into the file provider-id/name relative to the root directory of this repository.

Build the Images

Execute

docker-compose build

in the root directory of this repository.

Usage

Create and Start

To create and start the containers, run

docker-compose run -d

in the root directory of this repository. When started for the first time, the data-exchange service will generate a data provider-specific unique id which will be used for submitting quality indicator results without identifying information. A backup of this id should be made immediately since there is no way re-create or restore it if lost.

Stop

To stop the containers, run

docker-compose stop

in the root directory of this repository.

Start again after stopping

To start the containers again after stopping, run

docker-compose start

in the root directory of this repository.