Notice C

Promoting the collaborative development of proposals for investments in digital health global goods

Open Analytics on FHIR for Global Goods

Notice C Opportunity: 
Announcement C0: Global Good Software Development and Support
Application Status: 
Incomplete

I. Executive Summary

Current global digital health goods like DHIS2, OpenMRS, and iHRIS have their own siloed reporting systems, which are functional for viewing each system’s data independently but prevent comprehensive reporting and data analysis across systems. In addition, software developers need to continuously recreate specific dashboards for each system as users seek new ways to analyze data. This inability to create unified data analysis and dashboards impacts all levels of health care systems.

Open standards like the FHIR (Fast Healthcare Interoperability Resources) specification for exchanging health care information make is possible to develop health system analytics and dashboards that can be written once and run anywhere. There is support for FHIR in OpenMRS, iHRIS, and OpenInfoMan, and we can use their data based on FHIR in reusable applications across these and other global digital health goods.

Open Analytics on FHIR for Global Goods will create open source, shareable analytics tools for patient and health workforce data in the FHIR format that will strengthen and support existing global goods. The interactive “notebooks,” based on Jupyter Notebooks, will enable software developers and even non-technical staff to analyze FHIR data from multiple sources, such as patient records, health worker registries, and aggregated health trends, in one easy and convenient format.

II. Consortium Team

IntraHealth International has a long history in developing successful data tools for digital health applications. From mobile apps to management software to multi-language interactive voice response, we offer health workers and managers the tools and technology they need to do their very best work.

We develop solutions that are open source, data-driven, sustainable, and collaborative. And as a pioneer in the field of health workforce informatics, we’re committed to using technology, information, and analytical approaches to support the people at the center of our health systems.

This intervention will be led by the following IntraHealth staff and supported by a full range of health experts, project managers, and software developers.

  • Richard Stanley, Digital Health Product Manager, has over 20 years of experience in information and communication technologies, including high quality research and rapid data analytics for monitoring and evaluations in Afghanistan, Somalia, South Sudan, Uganda, and Sudan. He holds a PhD from the University of Oxford, UK.
  • Luke Duncan, Digital Health Assistant Director, has over 20 years of experience in software development, including leading the developing of iHIRS, the flagship human resources solution for global health, and multiple data interoperability standards and reference designs to connect iHRIS, DHIS2, and OpenMRS.

DAI combines 40-plus years of experience in global health with the latest data management and visualization tools to deliver comprehensive health solutions while responding to issues ranging from emerging pandemic threats to HIV/AIDS to waterborne diseases.

IntraHealth International and DAI formed a strategic affiliation in 2017 to offer new combinations of integrated health and development services while retaining their respective name, distinct identity, and legal status. DAI will directly contribute to this effort with the following staff:

  • Greg Maly, Principle Data Scientist, has over 13 years experience in data management, analysis, and visualization with R, Python, and QGIS, and trained international development staff on data analysis and geospatial technologies.

III. Project Description

Problem Statement

Current global digital health goods like DHIS2, OpenMRS, and iHRIS have their own siloed reporting systems, which are functional for viewing each system’s data independently but prevent comprehensive reporting and data analysis across systems. In addition, software developers need to continuously recreate specific dashboards for each system as users seek new ways to analyze data. This inability to create unified data analysis and dashboards impacts all levels of health care systems:

  • Clinic staff cannot see a comprehensive view of how their specific efforts impact national indicators.
  • Health care managers cannot link staffing levels with patient activity or health outcomes.
  • National policy-makers are frustrated by siloed data and manual workarounds to produce key reports.
  • Software developers waste valuable resources creating parallel systems to generate needed data flows and customizing dashboards for multiple systems.

The net result is overall disappointment in health data systems and a major barrier to expanding the culture of data-based decision-making across ministries of health.

Open standards like the FHIR (Fast Healthcare Interoperability Resources) specification for exchanging health care information make is possible to develop health system analytics and dashboards that can be written once and run anywhere. There is support for FHIR in OpenMRS, iHRIS, and OpenInfoMan, and we can use their data based on FHIR in reusable applications across these and other global digital health goods.

DHIS2, OpenMRS, iHRIS, and OpenInfoMan are existing global goods as they are deployed in multiple low and middle income countries, in the health sector, and are made available under Open Source Initiative approved software licenses.

Although FHIR is an open standard, there are only a few free or open source analytics tools available. Most FHIR analytics tools are proprietary, sold for use inside specific software and packaged for commercial use as FHIR is a new and complicated standard. Additionally, these analytics tools lack privacy protections required when using patient-level health care data.

Open Analytics on FHIR for Global Goods will create open source, shareable analytics tools for patient and health workforce data in the FHIR format that will strengthen and support existing global goods. The interactive “notebooks,” based on Jupyter Notebooks, will enable software developers and even non-technical staff to analyze FHIR data from multiple sources, such as patient records, health worker registries, and aggregated health trends, in one easy and convenient format.

The data analysis tools will also include processes to remove user-owned content and personally identifiable information to meet both required and expected privacy protections, and allow for users to experiment with introductory machine learning tasks currently beyond the abilities of existing global digital health goods.

Technical Approach

We will start this project with a detailed assessment of the current state of DHIS2, OpenMRS, and iHRIS compliance with FHIR. At this time:

  • DHIS2 doesn't support FHIR yet, but there is strong interest in supporting FHIR for indicators, facilities, and administrative units.
  • OpenInfoMan can be used to export facility and provider-level data.
  • OpenMRS has comprehensive support for FHIR import and export.
  • iHRIS supports FHIR data export.
  • OpenInfoMan has core support for FHIR export.

Then we will engage the global health community to develop data use cases and reporting needs across all three global goods to develop initial reporting needs and dashboards.

With a representative set of use cases, we’ll use the Python software language, which emphasizes code readability, and iterative agile software development processes to develop analytical tools like Jupyter Notebook, an open source web application that can contain live code, equations, visualizations, and narrative text.

The data analysis tools will include functionality like data anonymization, cleaning and transformation, numerical simulation, statistical modeling, data visualization, and machine learning. FHIR data can also be exported from the tools in CSV and similar formats for importing in DHIS2, Tableau, PowerBI, and other analytical tools.

We will engage the global health community to help us test and debug the data analysis tools, with a focus on software developers and policy-makers in low- and middle-income countries. Our goal will be an initial set of data analysis tools that are immediately relevant and useful to health data system managers, and provide motivation for key stakeholders to develop their own FHIR-based reports and dashboards.

We’ll support this initial interest with extensive documentation, including tutorials on FHIR profiles, FHIR servers, visualization production, and basic machine learning, to lower barriers of entry to producing other data analysis tools. We’ll also launch a targeted marketing campaign to publicize the availability of the data analysis tools and the ease of creating new ones using approaches like the Jupyter Notebook framework.

Anticipated Timeline

Month 0-2: Phase 1 - FHIR and Report Research. Confirm the FHIR standard compatibility of DHIS2, OpenMRS, and iHRIS, and the ability of OpenInfoMan and other tools to create FHIR-compliant data streams, and develop the initial needs and use cases for data analysis and reporting. Deliverable: initial set of application data flows and reporting use cases.

Month 3-6: Phase 2 - Iterative Prototype Build. Using iterative Agile software development processes and following Open Source software best practices, develop prototype interactive data analysis tools and reporting dashboards. Deliverable: beta data analysis tools published to Github.

Month 7-8: Phase 3 - Initial Large-Scale Testing and Debugging. Engage the greater global health community in a thorough testing of the data analysis tools and the reporting dashboards they produce to ensure usable analytics. Deliverable: user-tested data analysis tools suitable for public use.

Month 9: Phase 4 - Documentation and Marketing. Publish documentation, including tutorials on FHIR profiles, FHIR servers, visualization production, and basic machine learning, and launch a targeted marketing campaign to publicize the availability of the data analysis tools to the global health community. Deliverable: usable documentation and an executed marketing campaign.

Monitoring and Evaluation

IntraHealth International and DAI have robust monitoring and evaluation processes to ensure project compliance and success.

We will start this engagement with a deep discussion with representative software developers and policy-makers who will form an advisory group to confirm our initial needs assessment and create a clear future vision, with documented success criteria.

As we proceed through the data analysis tool development process, we will monitor our progress with regular check-ins with the advisory group to make sure that we are still building toward our future vision. We will also bring in other global health community members to ensure our data analysis tools have the greatest overall utility.

Once the data analysis tools are developed and we begin publicizing their utility, we'll evaluate our overall efforts to measure how well we've met our initial objectives and the extent to which the new tools are changing the way software developers and policy-makers approach digital health data.

IV. Tagging

  • Analytics
  • Interoperability
  • FHIR
  • Data Analysis
  • Data
  • OpenMRS
  • iHRIS
  • DHIS2