Other · Fhir Omop Etl Pipeline
Homa health
Healthcare Analytics & Contract Intelligence SaaS platform.
The brief
Design and build a comprehensive employer health plan analytics platform enabling health plan administrators and benefits consultants to monitor member spending, identify clinical risk, manage contracts, and surface actionable AI-powered insights — starting with product discovery and UX design in early 2024 before transitioning to core data engineering in 2025.
What we built
A dual-layer health analytics platform for employer-sponsored health plans. Layer 1 (UX/Product): High-fidelity designs for onboarding, workflow management, playbooks, contract AIQ, document comparison, risk register, and an experience/insights screen — delivered across two design phases (early 2024 and early 2025). Layer 2 (Data Engineering): A production FHIR→OMOP ingestion and transformation pipeline for multiple employer groups (Hero Health, DCC, and others), feeding a claims analysis engine (surge, emerging, top spending), a 19+-KPI clinical/financial library (including NYU-ED algorithm), a 6-dimension patient cohort engine (diseases, procedures, drugs, demographics, facilities, providers), and a MongoDB-backed insights schema — all deployed across dev and prod VMs with automated batch orchestration, pre-commit code quality hooks, uv-managed Python environments, and Drata compliance controls.
FHIR→OMOP ingestion pipeline running in production for Hero Health and DCC employer groups; claims analysis data (surge, emerging, top spending) live in MongoDB for all groups; 19+ KPIs with parameterized default/master lists deployed to prod VM; 6-dimension cohort engine integrated into analysis pipelines; dev and prod VM infrastructure with automated batch drivers, notification services, and GitHub-managed repositories; UX designs for full product suite (onboarding, dashboard, reports, playbooks, contract AIQ, risk register, document comparison.
Delivery timeline
How it was built, phase by phase.
8 workstreams across 88 weeks of operated delivery.
- discoverWeek 1–6 (Jan 22 – Feb 29, 2024)
Discovery, Research & UX Strategy
Initial engagement phase covering stakeholder interviews, competitive analysis of health tech platforms, HMW (How Might We) activities, affinity diagrams, user journey mapping.
Competitive landscape, HMW artifacts, user journey maps, and information architecture defining the product's onboarding, dashboard, AI insights.
FigmaFigjamMiro - designWeek 3–7 (Feb 6 – Feb 29, 2024)
Onboarding & Workflow UX Design
End-to-end UX design of onboarding flows, workflow management, task management flows, and playbook edit flows — iterated extensively based on stakeholder feedback.
Wireframes and high-fidelity prototype covering onboarding, workflow, task management, playbook editor, and document management modules.
Figma - designWeek 5–8 (Feb 15 – Mar 7, 2024)
Analytics Dashboard, KPI & Report Hi-Fi Design
High-fidelity design of report tabs including KPIs, rates, tables, and analytics dashboards.
Fully prototyped high-fidelity report suite with KPI widgets, rate tables, and contract view designed for health plan analytics consumers.
Figma - buildWeek 55–80 (Feb 2025 – May 2025)
FHIR-to-OMOP Data Ingestion Pipeline
Core healthcare data engineering work — ingesting raw FHIR-format claims data (medical, Rx, census) from multiple employer groups and converting it into the OMOP CDM standard in both dev and prod environments.
Multi-group FHIR→OMOP ingestion pipeline operational for Hero Health and DCC employer groups covering medical, Rx, and census data streams.
PythonFHIROMOP CDMParquetPostgreSQLSFTP - buildWeek 56–68 (Feb – Apr 2025)
Claims Data Transformation & Quality Validation
Systematic transformation of ingested claims data — applying business rules, date filters, episode-of-care logic, stop-loss aggregation, waste calculation.
Validated transformation pipeline with cost reconciliation logic across all employer groups, with configurable date filters, rules engine.
PythonSQLPandasPolarsParquetPostgreSQL - buildWeek 70–84 (May – Aug 2025)
Claims Analysis Engine (Surge, Emerging & Top Spending)
Development of a multi-dimensional spending analysis framework covering surge-in-spending, emerging-spending, and top-spending analyses across employer health groups, with cohort tagging, MongoDB persistence.
Modular claims analysis framework with three spending analysis types, dynamic time periods, cohort classification (Med/Rx).
PythonMongoDBPandasPolarsJupyter Notebooks - buildWeek 75–88 (Apr – Sep 2025)
Healthcare KPI Framework (19+ KPIs)
Design and implementation of a comprehensive KPI library (19+ KPIs) covering clinical and financial health metrics — including NYU-ED algorithm, ICD-level coding, chronic disease cohorts.
19+ health KPIs with parameterized UIDs, master/default list schemas, dynamic time period support, Rx edge-case handling.
PythonMongoDBSQLICD codesHCPCSNYU-ED Algorithm - buildWeek 76–82 (May – Sep 2025)
Patient Cohort Segmentation Engine
Development of multi-dimensional patient cohort filters across chronic diseases, facilities, procedures, drugs (NDC/RxNorm), demographics.
Six-dimension cohort engine (diseases, facilities, procedures, drugs, demographics, providers) with null-value handling, test coverage.
PythonICD codesHCPCSNDC codesSQLMongoDB
More case studies
Related work
09 · Run a function
Stop renting hours. Start running functions.
Pick the function you want off your plate. We'll map the brain and name the outcome we'd commit to — before you do.