Implemented CDC feeds, coverage and sample inputs

FluTrack's implemented index inputs are NSSP emergency-department visits, NSSP Acute Respiratory Illness activity and NWSS wastewater from data.cdc.gov, read through the existing ingestion service. Up to three usable signals contribute to a live combined respiratory index; NREVSS laboratory positivity has no live adapter. The model can include positivity only in labeled illustrative samples. Our methodology describes the unchanged weights and thresholds.

Index feeds, separate measures and unavailable inputs

Each source above is a weekly time series, not a live count. Surveillance data is reported with an inherent lag of roughly one to two weeks. FluTrack shows directional trends, not a real-time case count. A supported trend compares the latest observation with the mean of up to three prior observations, rather than an ordinary week-over-week change.

How the data reaches your screen

FluTrack works in three tiers, with provenance labels indicating whether a reading is observed or illustrative:

• Built from the CDC feeds. A build attempts the implemented index feeds through the approved caching service and uses observations that pass coverage checks. A live-origin badge identifies reported inputs; their observation periods and age determine what they can support, not the build time.
• A sample fallback. If usable observations cannot be obtained, a bundled illustrative snapshot can keep the interface usable. Those readings say Sample data; any accompanying explanation describes an example, not actual state conditions.
• A refresh in your browser. When the page you loaded holds sample data, or a newer CDC week may have been published since it was built, your browser fetches the same feeds and updates the page.

The “Check your area” card and /api/official

The county card reads a separate copy of the CDC data kept by FluTrack's own ingest service, which queries data.cdc.gov directly every Wednesday and Friday afternoon (UTC). Each dataset is pulled, checked and stored on its own: one that fails, or returns too little to be a real weekly release, changes nothing, and the last good week stays in place with its own date. Missing data is shown as missing, never as zero, and anything more than 14 days past its week ending is flagged. The same official-only data is available as JSON at /api/official?state=CA&county=06073 (open to any site, with attribution to the CDC).

Two further sources are built but switched off for now: the CDC's influenza A wastewater sample data (ymmh-divb), whose WastewaterSCAN rows are always dropped, and state influenza-like-illness rates from CDC ILINet as re-served by Carnegie Mellon's Delphi Epidata API. The CDC's FluSight forecasts, NREVSS regional positivity and California's CDPH respiratory dashboard are candidates for later.

Both the build and browser use the caching copy at ingest.oakandmain.dev. Cache retrieval and source publication follow different schedules. An October retrieval of a September observation remains a September observation. Each metric retains its geography, observation period, publication date when supplied, retrieval timestamp and usable coverage; an absent source publication date is unknown rather than inferred from the build. Source fields named buildnumber or date_updated remain source build/update dates, not verified publication dates.

The one exception is the CDC's epidemic-trend file. It is not a Socrata dataset and is not on the ingestion service, so the build fetches it directly from cdc.gov and checks its schema, dates and categories before using it. Your browser never fetches it. If it cannot be fetched or fails those checks, the epidemic-trend blocks are left out; they are never replaced by sample values, and pages built from sample data carry none. The figures behind every block are published as /data/epidemic-trends.json. How FluTrack shows it

Coverage and interpretation

Cases, laboratory positivity, wastewater viral activity, emergency-department visit percentages and hospital admission rates measure different things. Hospitalization data is displayed separately from the combined index. State observations do not independently measure a city, and participating sewersheds do not cover every resident. No data does not mean no illness. A fresh cache timestamp or a list of 51 jurisdiction keys does not establish usable observations for each pathogen or source.

Weekly releases can be delayed or revised, and preliminary and final hospital products have different publication schedules. A missing week, insufficient trend history or stale observation is qualified rather than converted to zero or a stable trend. For current official context, see the CDC's respiratory data, activity levels and state wastewater pages. National trends and local changes cannot substitute for a selected state's observations.

Documented CDC surveillance gaps — October 2, 2026

• Wastewater contract transition. The CDC's state wastewater page, updated October 2, 2026, says a new testing contract was awarded to Verily on September 28, 2026. CDC warned of a brief reporting gap affecting about 200 sites while sampling, testing and reporting restart. This notice alone does not establish why a September 26 observation is missing; each reading's own dates and coverage still apply.
• Iowa NSSP reporting. The CDC's activity-level data notes, updated October 2, 2026, say Iowa's NSSP feed ended on May 6, 2026 following a change in health information exchange vendors. Only data through the week ending May 2 is considered; weeks ending May 9, 2026 and later show Data Unavailable. This reporting gap is not evidence of low illness activity.
• Missouri's ARI baseline. The same October 2 CDC notes explain that data quality issues prevented use of Missouri's own historical data before MMWR Week 10 of 2025 (March 2–8). CDC used the rest of HHS Region 7 for that earlier portion of the historical baseline, and Missouri's own data from Week 10 onward, to calculate its 2026–27 activity levels. This is a caveat about CDC's categorical baseline, separate from FluTrack's editorial index thresholds.

Documented CDC surveillance gaps — October 9, 2026

• South Dakota emergency-department data. The CDC's activity-level data page, updated October 9, 2026, says no emergency-department visit data are available for South Dakota. The CDC estimates its epidemic trend from emergency-department visits, so FluTrack withholds that category for South Dakota, and for Iowa for the reason above, rather than show a direction label without the data it rests on. This reporting gap is not evidence of low illness activity.
• No CDC trend estimate. The CDC does not estimate an epidemic trend where emergency-department data are too sparse, show recent anomalies, or fail its reliability checks. FluTrack says so in those cases; “not estimated” is not the same as “not changing”.

Licensing and what we deliberately exclude

FluTrack uses only public-domain U.S. Government data — the CDC's own surveillance products, which carry no usage restrictions and can be reused by anyone, including on a commercial site. That constraint is a deliberate design choice, not an accident of what was easy to find.

FluTrack deliberately excludes WastewaterSCAN / SCAN data licensed CC BY-NC 4.0, which permits non-commercial use only. FluTrack is designed for commercial use, so those inputs are excluded. A provider name is not a license: the CDC's September 28, 2026 Verily testing-contract award is a distinct source context and does not automatically identify the resulting data as WastewaterSCAN data.

Independent, not official

Building on the CDC's open data does not make FluTrack a government product. We read the same public feeds available to everyone, with no affiliation, funding relationship, or special access.