Using FHIR on Android for standards-based care in Liberia Peter Lubell-Doughtie HL7 FHIR DevDays June 7, 2023

Outline 1. Liberia project impact 2. Mobile healthcare on FHIR 3. How we apply FHIR tooling 4. Next steps

Reduce the mortality and improve the livelihoods of children under 5 ● Register households ● Reproductive, maternal, and neonatal health ○ e.g antenatal care ● Child health ○ e.g. childhood vaccination ● Record health commodity supplies ● Tracking and supervision of health workers In 2021 under-5 deaths per 1,000 live births were 4 in the Netherlands, 39 in Rwanda Project scope: delivering high quality care to Liberians

2,000,000 patients cared for by 4,000 healthcare workers managed by 500 healthcare supervisors People spread over 100,000 km2 ○ ~3x size of the Netherlands with ¼ of the people Personalized healthcare powered by digital platforms

The vast majority of the population in need of healthcare services have no mobile connectivity and are multiple days travel from an internet connection Mobile network coverage in Liberia Built to manage care offline

Improving healthcare in Liberia

FHIR Store + API Extensions Analytics Manage access and extend the FHIR API with required features for global health Dashboard and dataware house, integrations to national data systems, e.g. LMIS / DHIS2 Web app for user, data and content management Off-line first app with dynamic content Admin Web App Identity & Authz HIS Integrations OHS: Android FHIR SDK Keycloak Airbyte, Superset A mobile platform built on FHIR HAPI FHIR

FHIR-based platforms FHIR-based app logic Standardized data model Software that shares components like Legos FHIR digital health architecture = building blocks

FHIR resources define project Health workflows: Questionnaires , PlanDefinitions , StructureMaps Transactional data: Patients, Observations, Tasks UX configuration: Profiles, Registers, Menus Look and feel State machines and forms Data viewing FHIRPath CQL Evaluator SDC FHIRPath SDC Android FHIR SDK Data Functionality Tooling Structure of a FHIR native app

Data: OpenSRP ANC app configs Tooling: FHIRPath queries on codes Functionality: Register & Profile views Configs and FHIRPath control UX

WHO Implementation Guide Local adaptation OpenSRP app configs co-develop with health officials and experts Global standards adapted to local context

Trigger the generation of a care schedule represented by a CarePlan and Tasks from a PlanDefinition E.g. Preeclampsia form submitted indicating risk factors are present Use color coding to guide health worker on upcoming, due, overdue and expired Tasks Questionnaire submission Schedule and Task engine Display care activity schedules CarePlan containing generated Tasks Health workflows through PlanDefinitions and Tasks

Patient Encounters Condition Observations Locations Tasks Groups FHIR Resources CQL Evaluator Android FHIR SDK $evaluate-measure Measures are written in CQL and encoded and represented as ELM libraries MeasureReports are generated and stored on the device with data from them shown to the user OpenSRP using the Android FHIR SDK and CQL Evaluator libraries generates MeasureReports through the FHIR $evaluate-measure operation Quality Measures (QMs) are represented using the FHIR Clinical Reasoning Module in Clinical Quality Language (CQL) Quality reporting with measures

Tracking stock with Questionnaires and measures Healthcare worker submits a Questionnaire with data indicating they used an RDT A StructureMap process the QuestionnaireResponse to create a stock-related Observation The configuration defining the list of commodities uses FHIRPath to aggregate Observations and calculate the current balance View in list of commodities View as historical reports Viewing reports triggers the $evaluate-measure operation to aggregate resources over a fixed time period for all commodities

Tracking stock with Questionnaires and measures

Clinical reasoning through structured data capture Enabled by SDC expressions using FHIRPath enableWhenExpression ● supports skip logic initialExpression ● to load default values answerExpression ● to load existing data calculatedExpression ● to calculate responses based on data entered so far

● 2 pilots complete — focus on national rollout in Q3 ● Collaborate with FHIR community on use-case specific representations Next steps and future work ● Data driven health interventions ● Personalized team-based care

Get in touch! [email protected] [email protected] Powered by our partners and the community