Methodology
How data is collected, deduplicated and verified.
Data pipeline
Case features are fetched on demand from the upstream ArcGIS FeatureService, cached in Cloudflare KV with a 5-minute TTL, and served to the map client via an edge-optimized JSON endpoint. The pipeline degrades gracefully: if the upstream is unreachable, the map renders the last cached snapshot rather than failing.
News headlines are aggregated from Google News RSS across four targeted queries (hantavirus, ANDV hantavirus, MV Hondius hantavirus, hantavirus outbreak), deduplicated by URL, sorted by publication date and capped at 60 items.
Geocoding and bucketing
Case locations are mapped to ISO-2 country codes via a curated lookup table covering major cities and regions in active outbreak countries. Cases that cannot be confidently assigned to a country (such as shipboard cases attributed to MV Hondius) are bucketed under an INTL pseudo-country and surface via the appropriate cluster page rather than a country page.
Frequently asked questions
Where does the case data come from?+
Primary case data is pulled from the public ArcGIS FeatureService Tracking_Hantavirus_2026 maintained by independent researchers, cross-checked against WHO Disease Outbreak News bulletins, national health agency epidemiological reports (Argentina, Chile, Switzerland, USA, etc.), and verified open reporting aggregated via GDELT.
How often is the map updated?+
Cases are refetched every 5 minutes from the ArcGIS source through a Cloudflare KV cache. News headlines are refreshed every 15 minutes from Google News RSS. The Last update timestamp in the map header reflects the most recent successful fetch.
How are duplicates handled?+
Cases are deduplicated on the upstream feature service by ObjectID. News items are deduplicated by URL and title before display. We do not merge cases across sources automatically — if a case appears with conflicting metadata across reports we prefer the most specific official source.
What counts as 'confirmed' vs 'suspected'?+
We mirror the status assigned upstream. Generally, confirmed cases have laboratory confirmation (RT-PCR or serology), suspected cases have a compatible clinical and epidemiological picture without lab confirmation at the time of reporting. Status can change as testing completes.
How accurate are the case locations?+
Case points represent the patient's last documented exposure or place of diagnosis, not precise residential addresses. Coordinates are approximate, often at city or regional centroid level, and should not be used for individual identification.
Are person-to-person transmission chains modeled here?+
We tag cases by Exposure_Group and present clusters (e.g., MV Hondius, April 24th flight) as derived from upstream metadata. We do not run our own contact-tracing or genomic phylogeny — those are the responsibilities of national reference labs and WHO collaborating centers.