Production-grade spatial epidemiology, engineered for public health.
Reproducible, auditable, compliance-ready spatial analytics — with copy-ready, validated Python.

🌐 www.spatial-epidemiology.org

spatial-epidemiology.org is a practitioner's field guide to building disease-surveillance and public-health GIS pipelines the way an engineer would: deterministic data flows, explicit coordinate reference systems, privacy-preserving aggregation, and compliance by design. Every guide pairs the underlying statistics with runnable Python built on the open geospatial stack — geopandas, libpysal, esda, spreg, mgwr, pyproj, shapely, and scipy — with pinned versions, structured logging, and audit-ready provenance.

It is written for the people who ship this work in the real world:

• Epidemiologists & public-health analysts who need defensible cluster detection and access metrics
• Python / GIS developers automating surveillance and reporting pipelines
• Government and agency technology teams operating under HIPAA, GDPR, and state disclosure law

The audience is technical, so the writing is too: every page carries operational specificity — failure modes, edge cases, parameter-tuning rationale, and the compliance controls that make a result hold up under review.

The site is organized into four connected tracks, each moving from concept to validated implementation:

Coordinate reference systems for public health, spatial data types and formats, geocoding precision standards, HIPAA/GDPR compliance mapping, and spatial weights matrix construction — the adjacency object every downstream statistic silently depends on (Queen/Rook contiguity, distance-band, kNN, island handling).

Global & local Moran's I, Getis-Ord Gi* hotspot detection, Ripley's K point-pattern analysis, and space-time scan statistics — plus geographically weighted regression and spatial lag / error models for explaining why rates cluster, not just where.

Drive-time isochrones, batch OD routing with resilient error handling, facility capacity allocation, and spatial equity indices (2SFCA / E2SFCA) — including ESRI↔OSRM CRS reconciliation, route-matrix caching, capacitated load balancing, and distance-decay selection.

Geographic masking (donut and population-adaptive geomasking), spatial k-anonymity for microdata, dasymetric and small-count suppression for choropleth rate maps, and differential privacy for small-area aggregates (Laplace mechanism, privacy-budget allocation across nested geographies).

Every guide follows the same data-engineering anatomy — concept framing, a decision diagram, validated implementation code, edge cases and failure modes, and compliance / audit controls — with hand-authored inline diagrams, statistical notation in KaTeX, and structured data for rich search results.

The site is a static build produced by Eleventy:

npm install
npm run build     # generate icons, run code-block QA, and build to _site/
npm run dev       # local dev server with live reload

Content lives as Markdown under content/, with layouts and partials under _includes/, site data under _data/, and static assets under assets/. Diagrams are original, hand-authored inline SVG that adapt to light and dark themes.

Content © Spatial Epidemiology. The site code is published for reference and transparency.

Explore the full guide → www.spatial-epidemiology.org