How to use: Start on the Explorer page to filter and select a health center.
Click Health Center Profile to open that organization’s detailed profile in a new tab. Opening profiles in separate tabs makes it easy to compare multiple health centers side by side while keeping the main Explorer available for additional searches.
Interactive public-health analytics platform built from HRSA Uniform Data System data.
Stack: Google Cloud Run • Cloud Storage • BigQuery • Data Studio
Coverage: 2019–2025 • 1,300+ health centers • 17,000+ service locations
Architecture: HRSA UDS Workbooks + HRSA GIS API → Cloud Run ETL → Cloud Storage Raw Archive → BigQuery Raw & Analytics Layers → Data Studio
What I built: Automated ingestion of annual HRSA workbooks, metadata-driven schema mapping across changing source formats, dimensional and fact modeling in BigQuery, geographic enrichment using the HRSA GIS API, data-quality validation, and an interactive analytics application. The project was designed to handle meaningful year-to-year schema drift without relying on hard-coded source columns, while validation routines identify missing mappings, definition changes, duplicate keys, and unexpected source behavior before data reaches the reporting layer. The semantic model also standardizes EHR vendors into canonical vendors and vendor families and supports state and national peer benchmarking.
Key capabilities: Health-center discovery, longitudinal patient demographics, state and national clinical-quality benchmarking and percentiles, financial peer analysis, and EHR market-share and migration analysis.
This application uses publicly available HRSA Uniform Data System data from 2019–2025, supplemented with current service-site geography from HRSA’s public GIS service. Patient, quality, financial, and EHR measures are reported at the health center organization level; physical service locations are used for geographic exploration and should not be interpreted as independently reporting those measures.
Suppressed, withheld, or otherwise unavailable public values are preserved as unavailable rather than converted to zero. This distinction is particularly important for service and financial data, where a reported zero and a missing public value can have very different meanings.
Clinical-quality comparisons use state and national peer groups, with measure direction accounted for so higher percentile values consistently indicate stronger relative performance. Financial comparisons emphasize normalized measures and medians rather than treating larger raw dollar amounts as inherently better performance.
EHR market analysis uses standardized vendor and vendor-family mappings to reduce inconsistencies in source reporting. Historical market-share and switching trends describe the UDS health-center population and should not be interpreted as market share for the entire U.S. ambulatory EHR market.