SHIELD

Healthcare Access 2.0 — multi-scale care-delivery network analysis.

A bed is not access if the patient cannot reach, enter, afford, or continue through care.

SHIELD scores facilities, providers, service lines, and referral or transfer pathways across access, capacity, quality, workforce, affordability, continuity, equity, and surge resilience.

Series: Applied Systems

Infrastructure 2.0 family

SHIELD is one domain implementation of a shared evidence-first method:

PUBLIC SOURCES → CORPUS → SCORE → SERVICE PROMISE → GAP MAP
                                                     ↓
                                      CONCEPT → REVIEW → DESIGN

Lane	Repositories
Movement	ROUTE, GAUGE, TARMAC, HARBOR
Lifelines	PYLON, PACKET, BASIN, DRAIN
Public access	SHIELD, SLATE
Civic boundaries	ZONES

The family shares evidence labels, explicit scale and demand bases, T1–T4 service promises where meaningful, adversarial review, and acceptance of a rigorous null result. Each repository owns its domain semantics and safety boundary.

SHIELD is not a clinical study, licensing/accreditation determination, Certificate-of-Need or payer decision, medical advice, or advocacy brief, and it claims no CMS, state-agency, health-system, payer, or provider endorsement.

Use SHIELD

SHIELD is public and open to use as a reference model for aggregate, evidence-gated healthcare-access analysis. To scope a safe transfer test, source review, or aggregate-only local adaptation, start with docs/adoption/README.md.

Why this matters

Healthcare capacity is non-fungible: the wrong bed, clinician, specialty, coverage, referral path, or travel time cannot be repaired by a single total. SHIELD deliberately tests where physical-infrastructure scoring transfers—and where it breaks—on a service-and-human network.

First public-data result: the hospital footprint

SHIELD now reproduces the May 13, 2026 CMS Hospital General Information release as a bounded national denominator:

What CMS reports	Facilities
Medicare-registered hospitals	5,432
Flagged Emergency Services = Yes	4,498
Critical Access Hospitals	1,378
Rural Emergency Hospitals	41

The source contains 5,432 unique facility IDs across 56 states and included territories. Its hospital-type, emergency-service, and geography partitions each reconcile independently to the same denominator.

This is useful because it establishes where to start asking access questions. It does not answer them. Facility presence is not travel access; an emergency flag is not a staffing or response-time SLA; and unlike hospital types cannot be added as interchangeable capacity. Travel time, staffed beds, clinicians, service breadth, wait time, affordability, quality, outcomes, equity, need, costs, and savings all remain held.

County rurality distribution

The second public-data slice joins that footprint to USDA ERS 2023 Rural-Urban Continuum Codes without fuzzy matching or hand-written aliases:

Deterministic join result	Facilities
Matched to a RUCC county/county-equivalent	5,360
Metro county (RUCC 1–3)	3,456
Nonmetro county (RUCC 4–9)	1,904
Unmatched and left unallocated	72

Among matched facilities, 1,086 of 1,371 Critical Access Hospitals and 36 of 41 Rural Emergency Hospitals are in nonmetro counties. This describes county class, not distance, travel time, patient rurality, staffing, shortage, need, service quality, or adequacy. The 72 unmatched rows remain visible rather than being silently assigned.

Primary-care shortage registry

SHIELD also reproduces the July 31, 2026 HRSA primary-care HPSA download at its actual registry grain:

Current HRSA CSV result	Count
Designation-component rows	79,150
Unique currently Designated HPSA IDs	7,682
Unique Proposed For Withdrawal HPSA IDs	1,014
Unique Withdrawn HPSA IDs	8,999
Currently designated IDs spanning multiple components	762
Currently designated IDs spanning multiple rural-status values	282

HRSA's separate June 30 quarterly report counts 9,003 primary-care HPSA designations. SHIELD preserves that official total but does not force it to match the newer daily file: the sources have different dates and status surfaces. Component rows are not hospitals, counties, or people, and designation populations can overlap. This result therefore establishes a formal shortage-registry spine without assigning shortage, access, capacity, adequacy, costs, or savings to any hospital or community.

Designation–component–geography bridge

The same July 31 file now yields a geography bridge without flattening subcounty or facility designations into whole-county findings:

Current designation structure	HPSA IDs	Component rows	Multi-component IDs	Multi-county IDs
Single County components	2,088	2,248	89	89
Census Tract components	586	11,697	555	25
County Subdivision components	164	1,579	118	41
All area designations	2,838	15,524	762	155
Facility designations	4,844	4,844	0	0

All 15,524 area-component rows have internally consistent five-digit common county keys. Among facility designations, 4,826 do and 18 remain an explicit geography residual: 17 placeholder keys and one state-prefix inconsistency. Across both classes, 7,664 of 7,682 designation IDs have a validated common county key spanning 2,932 distinct codes.

This bridge locates designation components; it does not assign a designation to a CMS hospital, turn a tract or subdivision finding into whole-county shortage, deduplicate affected people, or establish access, staffed capacity, need, or adequacy.

Primary-care designation capacity formulas

The same current registry identifies exactly where HRSA records primary-care physician FTE and shortage values:

Formula coverage	HPSA IDs	Recorded FTE	Recorded shortage	Derived need met
Area designations	2,838	10,327.3034	11,498.4866	47.32%
Correctional facilities	550	308.1850	768.6050	28.62%
Capacity-bearing designations	3,388	10,635.4884	12,267.0916	46.44%

All 2,838 area records reproduce the served-population formula within half a person and the shortage formula within 0.01 FTE. Their aggregate population identity has a visible one-person rounding residual. The remaining 4,294 facility designations have no FTE or shortage value: 1,351 FQHCs, 171 FQHC Look-Alikes, 928 IHS/Tribal/Urban Indian organizations, nine Other Facilities, and 1,835 Rural Health Clinics.

These are designation-recorded primary-care physician quantities, not deduplicated people or clinicians. HRSA's formula excludes nurse-practitioner and physician-assistant services and specified automatic or service-based facility designations. The result is therefore a valid formula baseline, not a CMS hospital staffing, appointment-access, service-line, or adequacy result.

Why this is harder than physical infrastructure

SHIELD cannot treat capacity as a fungible physical flow. A staffed bed, specialist, clinic slot, payer pathway, referral route, transport option, and continuity relationship are not interchangeable.

That makes the evidence boundary stricter:

• use aggregate and synthetic fixtures unless a source is explicitly public and safe;
• never introduce patient records or individual medical recommendations;
• keep licensing, payer, Certificate-of-Need, accreditation, and clinical claims held unless a qualified external authority and source path support them;
• treat transfer-strain findings as service-network evidence, not medical advice.

What is implemented

Crate	Responsibility
shield-network	Care-delivery elements and pathway contracts.
shield-corpus	Evidence-labelled corpus validation.
shield-score	DIM-01..13 score artifacts.
shield-tier	Tier-SLA classification and shortfalls.
shield-gap	Gap analysis, transfer-strain evidence, and null results.
shield-cms-access	Reconciled CMS/USDA facility baselines, HRSA shortage-registry census, and held HLT handoffs.
shield-cli	Corpus, score, tier-SLA, and gap commands.

The implementation baseline is complete and fixture-backed. No patient records or clinical recommendations belong in this repository.

Quick start

cargo run -p shield-cli -- --help
cargo run -p shield-cli -- cms-access-baseline
cargo run -p shield-cli -- cms-access-held-pack
cargo run -p shield-cli -- cms-rurality-baseline
cargo run -p shield-cli -- cms-rurality-held-pack
cargo run -p shield-cli -- hrsa-primary-care-baseline
cargo run -p shield-cli -- hrsa-primary-care-held-pack
cargo run -p shield-cli -- hrsa-geography-baseline
cargo run -p shield-cli -- hrsa-geography-held-pack
cargo run -p shield-cli -- hrsa-capacity-baseline
cargo run -p shield-cli -- hrsa-capacity-held-pack
cargo test --workspace

Method

CORPUS -> SCORE -> TIER-SLA -> GAP -> CONCEPT -> REVIEW -> DESIGN

Documentation

• PRODUCT_PLAN.md
• docs/adoption/
• docs/vtrace/
• context/waves/
• .roles/ROLE.md

License

MIT. See LICENSE.