Skip to content

Methodology · Issue 04 · Revised Q1 2025

The signals, the lags, and the numbers we publish.

This page documents how FluTrack turns four independent data streams into a single influenza activity signal — what each stream measures, where it is loudest, where it is quiet, and how the fused output compares to public reference reporting across five consecutive seasons.

It is written for clinical epidemiologists, infection-control leadership, and the public-health analysts who review our work before deploying it. No marketing language, no aggregate promises — only the methods we use and the validation we have published.

DocumentFluTrack Methodology, v4.2 Last revised14 February 2025 AuthorsOffice of the Chief Epidemiologist, Boston, MA Peer reviewThe Lancet Digital Health, 2023 · 2024 update pending

Chapter I

Principles of signal fusion, and the boundaries of what we measure.

FluTrack is an aggregate surveillance instrument. We combine four independent signal streams — hospital clinical reporting, pharmacy co-purchase telemetry, wastewater surveillance, and traveler-clinic reporting overlaid on commercial air-traffic data — into a single normalized activity index that updates every six hours.

The platform is designed for situational awareness at the state, HHS-region, and national level. It is not a diagnostic tool. It does not produce patient-level case counts. It does not replace clinical judgment at the bedside, and it is not approved by FDA as a Class II medical device — the in-vitro diagnostic component of our wastewater assay workflow is currently in pilot review under the agency's breakthrough device designation pathway.

Three principles govern the work:

  1. Transparent signal fusion. Every published output can be decomposed back to its contributing streams. The fused index is reproducible from the input feeds and the published weighting schema.
  2. Lag-validated lead times. We measure detection dates against named public references (CDC FluView, HHS Protect, state DOH bulletins) and publish the deltas. We do not extrapolate beyond the seasons we have measured.
  3. Scope discipline. Seasonal influenza A and B are in scope. Novel pandemic strains, RSV, SARS-CoV-2 lineage tracking, and individual-patient case attribution are not.

Office of the Chief Epidemiologist · FluTrack, Inc.

Chapter II

Four signal streams, each with its own latency, coverage, and cadence.

No single stream is sufficient. The fused index is only as trustworthy as the slowest contributor and the thinnest coverage footprint in the panel. Below: the four streams, with their measurement windows and known blind spots.

01

Clinical encounter reporting

De-identified acute-care visits coded with influenza-like illness, A/B PCR confirmation, and admission flags.

Network
1,840+ acute-care facilities
U.S. admissions covered
71% of all admissions
Daily volume (peak)
3.2M de-identified encounters
Median latency
14 hours from encounter
Update cadence
Every 6 hours
Known blind spots
Outpatient-only networks, rural critical-access hospitals under 25 beds

Stream was the primary detector for the 2022/23 A(H3N2) early-season surge, sounding 11 days before HHS declared widespread activity.

02

Pharmacy co-purchase index

Aggregated, anonymized point-of-sale data on over-the-counter antipyretics and prescription antiviral co-purchases, normalized by store-traffic baseline.

Coverage
38,000 U.S. ZIP codes
Median latency
48 hours from purchase
Update cadence
Every 6 hours
Leading indicator
Detects new outbreaks up to 9 days before pediatric ED visits rise
Known blind spots
Insurance-mediated prescription capture; regions with low OTC pharmacy density

Used by the New York State Department of Health to set seasonal surge staffing thresholds in advance of clinical visitation data.

03

Wastewater surveillance panel

Sub-genomic quantitative PCR for influenza A and B at treatment-plant and sub-sewershed level, normalized by population-equivalent and flow rate.

Sites
412 across 38 states
Footprint
Largest privately operated influenza A/B signal in North America
Median latency
72 hours from sample
Update cadence
Every 12 hours
Validation
r = 0.87 wastewater-to-clinical correlation (The Lancet Digital Health, 2023)
Known blind spots
Industrial watersheds, tourist-season population flux, dilution events after heavy rainfall

Featured in The Lancet Digital Health (2023) for wastewater-to-clinical signal correlation; replication study pending 2025 publication.

04

Traveler-clinic & air-travel overlay

Geo-coded reporting from 192 country traveler-health clinics fused with OAG commercial air-traffic data to estimate inbound influenza exposure by destination.

Countries covered
192
Reporting sites
410 traveler-health clinics
Air-traffic overlay
OAG global schedule, refreshed weekly
Median latency
36 hours from encounter
Update cadence
Every 12 hours
Known blind spots
Land-border crossings, charter flights, regions with sparse clinic coverage

Field team in Geneva aligns this stream with WHO regional reporting conventions for cross-jurisdictional situational awareness.

All four streams are fused via a published weighted-average schema, versioned alongside the methodology document. Weights are reviewed quarterly by the 14-person Scientific Advisory Board.

Chapter III

Lead time vs. public reference reporting — measured, not promised.

The table below records the date on which the FluTrack fused signal crossed a defined activity threshold, alongside the date on which the named public reference crossed the same threshold, for the five most recent U.S. seasons. Lead time is the difference, in days, FluTrack earlier than the reference. Positive numbers are flu seasons where FluTrack led; negative numbers indicate the reference led.

Detection-date comparison — FluTrack fused signal vs. CDC FluView & HHS Protect
Season Dominant strain FluTrack detection Public reference Lead time Source
2019/20 A(H1N1)pdm09 11 Dec 2019 CDC FluView +6 days FluTrack archive
2020/21† B / minimal Not crossed CDC FluView — Suppressed season
2021/22 A(H3N2) 28 Nov 2021 HHS Protect +8 days FluTrack archive
2022/23 A(H3N2) 04 Oct 2022 HHS widespread declaration +11 days FluTrack / HHS
2023/24 A(H1N1)pdm09 19 Dec 2023 CDC FluView +4 days FluTrack archive
Median across five seasons (2019/20–2023/24, excluding 2020/21) 7.2 days FluTrack validation, 2024

† The 2020/21 season was suppressed by pandemic-era non-pharmaceutical interventions; no threshold crossing was registered by either system. Median lead time is calculated across the four remaining seasons. Lead times are validated retrospectively against public archives and are not extrapolated.

What this table does not claim.

  • We do not guarantee specific lead times for a given locale or season. Local wastewater coverage density, hospital network participation, and pharmacy-data availability materially affect detection speed.
  • We do not extrapolate beyond the five seasons measured. The 2024/25 season is in progress at the time of this revision.
  • We do not characterize FluTrack as a primary surveillance authority in parallel with CDC or WHO. FluTrack is a complementary, earlier signal — not a replacement.

Chapter IV

Four numbers, cited.

The minimum a clinical epidemiologist asks before evaluating a surveillance platform. Every figure below is published, peer-reviewed, or auditable on request.

r = 0.87

Wastewater-to-clinical signal correlation, influenza A/B pooled across 412 sites.

The Lancet Digital Health, 2023

7.2 days

Median lead time vs. CDC FluView, validated across four consecutive flu seasons.

FluTrack validation report, 2024

412

Wastewater surveillance sites across 38 states — the largest privately operated influenza A/B signal in North America.

Internal coverage audit, Dec 2024

1,840+

Acute-care facilities in the hospital-data network, covering 71% of U.S. admissions.

Network roster, Q4 2024

All four figures are reproduced in the printed methodology document and in our annual transparency report, available to qualified institutional reviewers under the standard data-use agreement.

Scope statement

FluTrack is a population-level surveillance instrument designed to support situational awareness for infection-control and public-health decision-makers. It is not a diagnostic tool. It does not produce individual case counts, does not assign risk to identifiable patients, and does not replace clinical judgment at the point of care. Outputs are intended to inform staffing, supply, and surge planning — not patient management.

FluTrack has not been cleared or approved by the U.S. Food and Drug Administration as a Class II medical device. The wastewater assay component is currently in pilot review. No content on this page should be construed as a guarantee of outbreak lead time for any specific locale, season, or pathogen.

Office of the Chief Epidemiologist
FluTrack, Inc. · Boston, Massachusetts

Have an analyst review our methods with your team?

Twenty minutes with our Chief Epidemiologist or a regional epidemiologist on your account — typically scoped to the four signal streams and your specific catchment area.

Request a demo

Or write to [email protected] · +1 (617) 555-0418 · 222 Berkeley Street, 14th Floor, Boston, MA 02116