Support Needs Reported
Share of reportable completed sessions that included each need.
Current Research Snapshot
Last 30 days
Earlier Research Sessions
—
Completed navigations collected under disclosure v1-2026-08-26. A separate, later round is counted on the Food & Nutrition tab — the two are never added together.
Top Support Need
—
No reportable demand yet
Top Barrier
—
No reportable barrier yet
Verified Resource Opens
—
Counted separately from intake answers
Multiple needs allowed
Barriers Reported
What participants say is getting in the way right now.
Potential Coverage Gaps
A need appears here when no strong verified match covered it in that research session. This flags a question to review, not proof that no service exists.
Geographic Activity
State totals from the aggregate counters, and five-digit ZIPs reported by Celestine participants once a ZIP reaches five.
Most-Opened Verified Resources
Observed engagement after the matching system returns verified options. Opens are counted separately from participant answers.
Current Demand
Observed research-participant needs and barriers, without predictive interpretation.
All Reportable Support Needs
Ranked by number of completed research sessions.
All Reportable Barriers
Ranked by number of completed research sessions.
Potential Coverage Gaps
Needs selected in research sessions that did not receive a strong verified match. See Potential Coverage Gaps for the supply context behind each one.
Geographic Research Signals
State totals and five-digit ZIPs, each shown only once the group reaches five.
Beta boundary: this page describes what research participants are asking for right now. Observed change between two equal periods is on the Food & Nutrition tab, delivered for the Data Exploration milestone. Predictive Analytics shows forecasts based on federal data for food assistance, housing, healthcare and employment in Florida, each with its Historical backtest performance. Celestine’s own session forecast publishes once enough weeks of sessions exist. Automatic recommendations and outcome tracking will come later.
Potential Coverage Gaps
Reported needs that did not receive a strong verified match, shown with available resource information for review.
How to read this: These are possible gaps, not proof that help is unavailable. A missing match may mean our resource information is incomplete, program rules or service areas are unclear, or no local service exists. Each item needs review.
How these numbers are calculated
So the numbers beside each gap can be checked.
Resource Engagement
What participants open after receiving verified support matches.
Verified Resource Opens
—
Reportable aggregate actions
Top Opened Resource
—
No reportable engagement yet
Reporting Threshold
5
Set by this dataset’s disclosure (v1-2026-08-26)
Join to Intake Answers
None
Resource opens are stored as separate counters
Resource Engagement Ranking
Only resources with at least five opens in the selected period are displayed.
Why This Signal Matters
A first behavioral signal beyond intake answers.
This separates reported need from observed action. Later, if people repeatedly report a need but do not engage with available matches, the team can investigate clarity, relevance, availability, or navigation friction. The Beta shows the behavior without making a causal claim or automatically recommending government action.
Architecture boundary: the server receives the verified resource ID by itself. It does not receive that participant's needs, barriers, ZIP, illness area, or a session identifier with the resource-open event.
Illustrative example
This shows how Resource Engagement could look with more activity. These are sample values, not participant activity.
Verified resource opens19
Top opened resource8example opens
Reporting threshold5
Connected to intake answersNo
| # | Example resource | Example opens |
|---|---|---|
| 1 | Food-support resource | 8 |
| 2 | Transportation resource | 6 |
| 3 | Housing resource | 5 |
Food & Nutrition
Nutrition demand observed in TOPx research, shown beside a separate seven-program research dataset. The two are never combined.
A
Nutrition Demand & Existing Assistance
Observed TOPx research dataFood Need Reported
Completed research sessions that selected food as a support need.
Food Need Without a Strong Match
Sessions where food remained unmet after matching. A question to review, not proof no service exists.
Health Coverage Reported
Health insurance categories across all reportable sessions in this period.
A2
Food Need, Benefits and the Kind of Support Asked For
Observed TOPx research dataDisclosure v2-2026-09-03Current Research Sessions
—
Completed navigations collected under disclosure v2-2026-09-03. The earlier round is counted separately on the Overview tab — the two are never added together.
Asked for food help
—
Sessions that selected food as a support need
Already receiving SNAP
—
Among those asking for food help
Still without a strong match
—
Among those asking for food help
What Kind of Food Support
Among sessions that asked for food help. Single-attribute counts, shown from one.
Assistance Already Received
What people report already receiving, across all sessions in this dataset.
Illness Area and Kind of Food Support
A two-attribute comparison. Shown only where a group reaches three.
Health Coverage Among Those Asking for Food Help
Relevant because some nutrition pathways require a particular coverage type.
Where Food Need Is Being Reported
Five-digit ZIPs where Celestine participants reported food need. A ZIP appears only once at least five sessions share it; smaller ZIPs are not named.
A3
USDA Food Access — County Reference Data
USDA reference data, not participant countsUSDA open dataUSDA Food Access by County — Reference Data
The counties where people told Celestine they need food support, shown beside what USDA measures about food access in those same counties.
B
Program Availability & Potential Access Barriers
Nutrition research datasetProgram Pathways
Seven researched candidate programs, grouped by the kind of help they provide.
Program Access Information
What each program publicly says about how people can access it.
C
Geographic Reach & Coverage Gaps
Nutrition research datasetCross-tab plannedWhere These Programs Reach
Coverage scope declared by each researched program, with its dominant entry condition.
D
Program Status & Verification
Nutrition research datasetStatus of Researched Programs
Why a live status field matters: a program can exist on paper and still be unavailable today.
E
Federal Investment in Food & Nutrition
Federal award dataWhere Federal Food and Nutrition Money Goes
Published federal award records for food and nutrition programs, shown at state level.
F
Reported Change
Observed Change in Reported Food Need
Compares food-need reports from two past periods.
This shows what changed between two past periods. It does not predict what happens next.
Geographic Activity
Where observed research activity is occurring. State totals come from the aggregate counters; five-digit ZIPs come from session records. Each appears only once the group reaches five. ZIPs are never added up into counties.
Observed Areas
States from the aggregate counters. Five-digit ZIPs from session records (disclosure v2-2026-09-03), where participants were told their ZIP is kept so organizations can see where needs are showing up. Earlier counter records never held a five-digit ZIP, so they contribute state totals only.
What This Cannot Yet Answer
Stated plainly so the limits of the current Beta are clear.
- Need by area — the counters hold need totals and area totals separately, not together.
- Barrier by area, condition by need, or coverage by need — same reason.
- Change over time by area — a trend needs a longer run of periods than this dataset has.
Predictive Analytics
Forecasts based on federal data.
Average miss is how far past forecasts landed from the actual figure, on average, as a percent of that figure. Celestine counts include team testing and real-user sessions.
Celestine’s own forecast
Weekly food-need sessions · Florida
Completed Celestine sessions that reported a food need. Not a count of people.
Loading…
Model card
About This Data
Research scope, privacy boundaries, architecture, and milestone pacing.
What This Dashboard Represents
TOPx focus: understanding support needs associated with invisible illness, including Lyme disease.
Aggregate activity from people using the TOPx research prototype. It is designed to show current observed demand and resource engagement during testing.
- Completed research-session counters
- Support-need categories
- Daily-life barrier categories
- Needs without a strong match
- State counters, and five-digit ZIPs from session records once a ZIP reaches five
- Verified-resource opens counted separately
What It Does Not Represent
This dashboard reflects people who used the TOPx research prototype. It does not represent population-wide community demand. Current Beta scope is observed demand, resource engagement, federal award context and observed change. Predictive Analytics shows forecasts based on federal data for food assistance, housing, healthcare and employment in Florida, each with its Historical backtest performance. Celestine’s own session forecast publishes once enough weeks of sessions exist. Automatic recommendations and outcome tracking will come later.
These results should not be described as overall community demand, disease prevalence, benefit eligibility rates, or a representative sample of people living with invisible illness.
- No forecasts about individuals — Predictive Analytics forecasts statewide federal figures
- No automated government recommendations
- No outcomes claims
- No individual-level conclusions
Aggregate-Only Reporting
The TOPx live-data bridge does not create an individual navigation row. Illness area, veteran answer, five-digit ZIP, matched-resource list, ranking score, free text, and contact information are not sent to this research aggregate table.
Groups under five are withheld for this dataset. The people in it were told that any group smaller than five would be withheld from the organization view, and that promise governs their records. This is a reporting policy recorded against the disclosure in force when the data was collected — not a fixed Celestine rule — and data collected under a later disclosure may use a different policy. Verified-resource opens are stored in separate counters, not joined to intake answers.
What Organizations Can Explore
The browser calls a same-origin Netlify function. That function validates an allowlist, then uses server-side Supabase credentials to atomically increment protected counters. The browser never receives the service-role key or direct table access.
Production hardening still includes formal organization authentication, rate limiting, audit metadata, retention policy, and stronger anti-replay controls.
Pipeline Verification
Useful for ongoing user research and validation without exposing small-group dashboard counts.
Waiting for a browser transmission
After this browser completes the navigator, this local status can confirm whether the aggregate request was accepted. It contains no intake answers.
Government Value Roadmap
Capabilities are deliberately paced across TOPx milestones rather than shown all at once.
Current demandNow / Beta
Observed changeData Exploration — delivered
ForecastsFederal data now — Celestine sessions as data grows
Recommended actionLater milestone
Outcomes → integrityLater roadmap
Methodology, sources and limits
The detail behind each tab, kept here rather than repeated on every card.
What Celestine’s session forecast needs
What Celestine’s Session Forecast Needs
The forecasts based on federal data run now. Celestine’s own forecast publishes when these conditions are met.
- Longitudinal history. Roughly 12–16 weekly observations per area-and-pathway series before a trend line means anything, and 26 or more before seasonality separates from trend. This dataset has a handful of sessions in total.
- Volume per cell. A forecast for one ZIP code needs enough sessions in that ZIP each week that the count is not noise. Sparse cells must be suppressed or aggregated up.
- A stable disclosure and taxonomy across the window. If need labels or geography granularity change mid-series, the history is not comparable.
- Resource-side history. Pressure is demand divided by supply, and supply has no time series yet.
- Scored forecasts. Celestine’s forecast publishes after 8 weeks of history and 8 scored live forecasts. The federal forecasts already show their Historical backtest performance.
A negative amount is not an error
A negative amount is not an error. Where obligations recorded in this period were reduced by more than they were increased, USAspending publishes a net deobligation. It is shown exactly as published rather than hidden or set to zero. : ;
How supply is counted
Verified supply is counted from the same catalogue the navigator matches against, not from a separate list. For each need, a resource counts toward supply when that need is in its matching profile. Reach is read from each record’s coverage scope: national and national-locator records are available in every state, the District of Columbia and Puerto Rico; state and state-network records are counted against the states observed in this reporting period; local records are counted separately because their reach is narrower than a state. The same architecture can support additional state pathways as partnerships and funding become available.
How these counts are produced
Demand and unmet counts come from aggregate counters and follow the same reporting threshold as every other figure on this dashboard. Supply counts come from the verified resource catalogue and are not participant data, so they are shown in full.
Why these counters are never added together
These are three independent counters. This dataset records each category on its own, so it cannot say how many people reporting food need also report a particular coverage type. Any such cross-tab would be a new capability, not a filter on this view.
Why these counters are never crossed
These three counters cannot be crossed
They come from the v1 counter dataset, which records each category on its own. Benefit participation and the kind of food support asked for are shown below, from the newer session-level dataset collected under disclosure v2-2026-09-03. The two datasets are never combined and never added together.
What this comparison is, and what it is not
What this comparison is, and what it is not
The county table is USDA’s published reference data only. Celestine participant ZIPs are never added up into counties; a county is listed only because it contains a five-digit ZIP where at least five participants reported food need, and that ZIP is named beside it. Each figure is USDA’s published measure for that whole county. Nothing here links any person to a USDA figure, and no individual record is used. A low-income-and-low-access flag describes a census tract — never a household, never a person, and never an eligibility decision. A county listed here means a reported ZIP sits in an area USDA measures as having limited food access. It is an observation worth following up, not a finding, not a cause, and not a forecast.
What a declared barrier does and does not mean
These are potential access barriers each program declares about itself. Celestine has not measured whether any person was actually stopped by one, and this view must not be read as though it had. unknown means the official source does not state it — it is never filled in by inference. Income limits are quoted from each program’s own notes; income is not a structured field in the access-gates schema.
Why demand and supply are not joined here
Supply side only. Observed participant demand by area is on the Geographic Activity tab and follows this dataset’s reporting threshold. The two cannot yet be joined: the counters record need and geography as separate one-dimensional totals, so “food need in this county against programs reachable from it” is a cross-analysis capability that has been designed but not built or approved. The county comparison that is built is in block A3 above, and it uses USDA’s published county measures rather than these seven researched programs.
What each status means
Verification dates are the day each record’s access details were last checked against an official source. unverified means the program exists but its current operating status could not be confirmed — it is not a claim that the program has stopped.
What this is, and what it is not
What this is, and what it is not
These are published federal award records, placed on the same page as reported need because an agency planning a service wants both in view. They are descriptive only. No amount here is divided by, multiplied with, or compared arithmetically against any Celestine session count, and nothing on this page says an area is underfunded, adequately funded, or funded in proportion to what people reported. Federal food assistance is largely awarded to state agencies, so a state total does not describe how much reached any one county, town or household, and a larger total does not mean a state is better served. Award records describe money obligated in a fiscal year — not services delivered, and not people helped.
What is designed but not approved
Each of these is a designed capability awaiting approval, not a defect. They are listed here so an agency reader knows exactly where the current architecture stops.
See the seven stages
1. Observed DemandLive
What people asked for help with, by category, at the reporting threshold.
2. Potential Access BarriersResearch data
Entry conditions programs declare about themselves. Exposure, never a measured outcome.
3. Geographic GapsLive, one dimension
Where activity is occurring. Area combined with need is designed, not built.
4. Trends — observed changeRunning today
The same count over two equal periods, subtracted. Reported on the Food & Nutrition tab. Observed only — never a projection.
5. ForecastFederal data
Where demand is heading, by area. Ranges, never point estimates. Forecasts based on federal data are on Predictive Analytics now; Celestine’s own session forecast follows as its data grows.
6. Recommended ActionNot built
A planning signal an agency can check by hand — never an instruction.
7. OutcomesNot built
Whether the help actually arrived. Nothing here is measured today.
The method, in full
Every figure in the method would be reproducible by hand: a least-squares fit for the trend, a band of plus or minus one standard error for the projection, and demand divided by available resources for pressure. No model an evaluator cannot check.
What outcome tracking would look like
Matched
Not measured
Verified options returned to the person. The nearest thing Celestine observes today is a resource open, counted separately from intake answers.Contacted
Not measured
The person reached the program.Barrier capture: could not reach, no response, wrong area
Applied
Not measured
An application or intake was started.Barrier capture: documents, referral, screening, wait list
Received help
Not measured
The service was actually delivered.Outcome capture: received, denied, still waiting
Why the funnel is shown at all
Celestine does not measure these outcomes today. Nothing on this dashboard is derived from them, and no figure anywhere else in the product depends on them. Showing the funnel is a statement of direction and of the barrier and outcome capture each stage would need — it is not a report. It is deliberately kept apart from observed demand and from declared access requirements, because gate exposure and a confirmed outcome are different things and must never be blended.