We propose creating a 100% FHIR native reference implementation of the WHO Digital Adaption Kit (DAK) for Antenatal Care (ANC). This will allow the application to be configured directly with L3 guidelines represented as FHIR resources, in the ANC DAK FHIR Implementation Guide. We will develop this using Google's Android FHIR SDK and OpenSRP FHIR Core - both of which we have been active in the development of over the past year.
The goal of the WHO SMART Guidelines is to help accelerate and systematize the consistent application of WHO-recommended guidelines for digital health applications. To help achieve this bold vision we believe a next generation of digital health applications that are FHIR native is needed. By “FHIR native app” we mean ones that are fully configurable using the machine readable machine readable guidelines (L3), as defined in the HL7 FHIR Implementation Guide for WHO’s ANC DAK.
With a FHIR native app you will be able to use L3 FHIR artifacts to:
represent personas and core data elements (mapped to terminology codes) as profiles of the appropriate FHIR resources.
define decision support logic using FHIR questionnaires, FHIR expressions and clinical query language (CQL)
define workflows and service delivery schedules using plan CarePlan definitions
define and calculate program indicators using MeasureReports
Since 2018, Ona has worked closely with WHO to build the first ANC reference app based on the guidelines that became the ANC DAK. We are currently working with WHO and local technical partners to adapt and implement this reference app in Malawi, Rwanda, Zambia, Indonesia and Nepal. The process of developing a reference app and then adapting it for new countries has given us a deep appreciation of the value that standards based FHIR configuration could provide especially for an app that requires more complex clinical decision support like ANC.
To be better positioned to support WHO Smart Guidelines we made the strategic decision at the start of this year to shift all of our digital health work moving forward to FHIR native apps. To this extent, we have been actively working with Google, WHO and a community of technical partners to lay the foundations for this by contributing to the development of Google’s Android FHIR SDK. Using this, we have been actively developing OpenSRP FHIR Core, a FHIR Native digital health platform, for supporting offline-capable, mobile-first, digital health applications. FHIR Core's user-friendly interface and modular design is based on over a decade of real world experience implementing digital health projects with OpenSRP. Using FHIR Core we have successfully developed a reference Covid-19 vaccine app for WHO and are currently developing a CHW platform that will be used in a national level roll-out by mid-2022.
We would use this investment from Digital Square to work towards implementing an initial version of a FHIR native ANC reference app using OpenSRP FHIR Core and Google’s Android FHIR SDK. Funding from Digital Square would help to augment significant internal investment Ona has made into FHIR Core and funding received from the Patrick J McGovern Foundation and WHO (initial seed funding) to lay the foundations for this work.
If we can achieve the goal of implementing the ANC reference app FHIR natively, we will make an important step towards the vision of the Smart Guidelines. In doing so, we hope to dramatically lower the barrier for local implementers to adhere to WHO guidelines in creating local adoptions while helping to retain a standardized data model that remains largely consistent across implementations.
The project will be led by Ona working in close collaboration with the Summit Institute for Development (SID) in Indonesia.
Ona is a global health technology solutions provider based from Nairobi, Kenya and Burlington, Vermont. Ona has over 70 employees and extensive experience designing, developing and implementing health information systems that are used at national scale and integrate with existing government health information systems and open standards. Ona has developed numerous open source standard tools and libraries in the space that are used by mobile teams to add functionality to their existing technologies.
For the past decade, Ona has served as the technical sponsor for OpenSRP, a global good digital health platform that we initially co-created with WHO and other partners. Ona has helped lead the development of digital health solutions that have been implemented in more than 22 countries, with OpenSRP-based implementations currently live in countries like Bangladesh, Chad, DRC, eSwatini, Guinea, Indonesia, Liberia, Malawi, Madagascar, Tunisia and Zambia - to name a few. These implementations primarily focus on the maternal and child health, childhood immunization and malaria disease surveillance and - management domains. In 2020 Ona has also started to build OpenSRP-based solutions to contribute to the fight against Covid-19 pandemic.
In terms of our collaboration with local tech and implementation partners in this space: The Summit Institute for Development (SID) is a global health organization based in Lombok, Indonesia. SID has deep expertise conducting large scale health implementation and research projects that involve the use of digital health technologies. SID was one of the founding members of OpenSRP, with Ona, and the two organizations have been working closely together for nearly a decade. Over the past two years, SID has invested significant efforts to develop FHIR expertise. This includes working with HL7 to develop a FHIR training programme adapted for global health groups. HL7 will also host a FHIR “dev days” in Indonesia next year, as a result of SID’s efforts towards advancing the FHIR-agenda in Indonesia and beyond.
Ona will serve as the technical lead for the project, but we will work closely with SID’s technical team to demonstrate the ability to adapt the reference app to the Indonesian local context. We are currently working to support SID to adapt the OpenSRP-based WHO ANC Reference app for Indonesia.
To achieve this work, will require the continued close collaboration with Google around the development of the Android FHIR SDK and other technical partners including technical experts from WHO who are contributing to the broader FHIR native app initiative.
Lastly, we have developed close working relationships with an expanding network of local technical partners. This includes BlueCode in Zambia, 1000Hills Solutions LTD in Rwanda, mPower Social in Bangladesh, VentureDrive in Pakistan and Softmed and D-Tree International in Tanzania who have all successfully developed and support the roll out of OpenSRP implementations in their respective countries. While all will not be directly involved in this initial effort, they are all committed to working with us to adopt FHIR Core on projects moving forward.
I. Background or problem statement
While we have made significant progress over the past year with partners like Google to develop FHIR native apps, there still requires some significant technical investment to achieve the full vision of a fully configurable FHIR app as defined by an HL7 FHIR Implementation Guide representation of the DAK. Currently, we can develop FHIR native apps where you can design data collection forms using FHIR questionnaires. Using data extraction we can create corresponding FHIR resources which can then be synced to a FHIR native server. Using these functionalities we have successfully built applications like a WHO Covid-19 Vaccine certification reference app. Also, through prior investment from PATH (through funding for Notice D: mADX on FHIR on Android) we are in the final stages of having added support to define and calculate program indicators in a FHIR app using MeasureReports and FHIR’s $measure-evaluate function. The key functionality required to implement a reference app as currently defined in the L3s, is support for complex clinical algorithms and decision support using CQL, and the ability to generate CarePlans and associated scheduled activities using the FHIR $apply function. Adding support for these two key missing functionalities will form the basis of the workstreams where we are requesting support and will significantly increase the shelf readiness of OpenSRP FHIR Core as a platform.
II. Objectives and activities
Below we describe 3 possible work packages, which we can choose amongst in collaboration with PATH and Digital Square based on the Global Goods community’s shared priorities. First, is adding support for ANC CarePlan generation which is a prerequisite required for ANC visit scheduling. Second, is adding CQL support for clinical decision support. Third, is working with SID Indonesia to develop a simplified version of the ANC reference app, localized to the Indonesian context for pilot testing.
III. Deliverables and Schedule
Work Package 1: ANC CarePlan definitions and activity generation required for ANC visit scheduling
First, we will work with WHO to define an ANC CarePlan definition and have the ability to configure it and load it into the ANC reference app. The CarePlan definition will be used to define the personas, workflows and scheduled activities associated with an antenatal care plan. Second, by leveraging the $apply function we will add the ability to use the CarePlan definition to schedule ANC visits as required by the ANC DAK. We will also do the required work for FHIR Core to associate these tasks as scheduled events.
Work Package 2: Complex clinical algorithms and decision support logic using CQL
First, we will work with Summit Institute Indonesia, to help ensure the clinical guidelines (as represented in FHIR Questionnaires and associated CQL statements) are in a format that can be loaded into the FHIR Core app for implementation. We will then seek to add support for Clinical Quality Language (CQL) into FHIR Core. CQL will make it easier to incorporate advanced clinical decision support (as defined in the L3s) in addition to being able to evaluate triggers in CarePlans - that can be used to identify red flags that indicate a woman is at risk and recommend an associated care intervention.
Work Package 3: Simple Reference ANC App localized to Indonesian Context
We will work with SID to develop a simplified ANC reference app that is localized for the Indonesian context. By simplified, we will aim to support the key workflows, clinical decision support and data model defined in the ANC DAK. Depending on the maturity of the L3 definitions (see below) and Indonesia’s needs, it is a real possibilityble that not all content and workflows will be supported through this grant. Lastly,In addition, due to the budget ceiling of this award, constraints we will not be able to develop all of the functionality needed to support all of the functional and non-functional requirements defined in the DAK, but we are open to agreeing to achievable priorities (within the agreed-upon timeline and budget) with Digital Square and the broader community..
Depending on when the contract is awarded, we foresee to complete the required technical work to over a span of six months for the initial reference app. Would would then need about 3 months to develop with SID the local adaption and then allow 3 months for them to locally test.
IV. Risk Mitigation
There are a number of external risks associated with the development of the Android FHIR SDK that will affect our ability to deliver this project. For we are dependent on the FHIR SDK to support both CQL and the $apply function before we can integrate it into FHIR core and convert to a working app. We have been working closely with Google and WHO technical partners to ensure these will be in place hopefully over the next two months. There are also a number of implementation decisions in FHIR like the best way to define activities in a PlanDefinition that still need to be researched and decided upon in our FHIR implementation community.
Second, while a lot of the required L3 FHIR resources have been developed, the work by WHO is not complete. Implementing the full ANC DAK will require these to be in place. This may take some time and back and forth with WHO to complete, and although we don't have objections against the process to be followed, we are cognisant of the time that this will take to complete.
Third, there are currently no official Implementation Guides (IGs) in place for a FHIR based ANC Reference Applications. Our suggestion is that we work with the broader “digital health applications for ANC”-community, to agree on, and come up with, an IG for this ANC POC. If we do not do that, each mobile application could effectively decide on their own approach for how they want to implement the SMART Guidelines and the DAK. This fragmentation would make downstream “apples to apples”-comparison and analysis very difficult. We are proposing to come up with a common IG, so that this ANC POC can be tested and implemented in a standardised manner.