Hypersphere Core Platform v15 — Hypersphere Health™

📋

CLINICIAN WORKFLOW GUIDE — HPDM™ EIE™ v10.1

Pre-Encounter Intelligence · Attestation Protocol · RADV Compliance · Click to expand / collapse

42 CFR § 422.504(l) ▼

WHAT IS THIS TOOL?

The HPDM™ Patient Engine is a pre-encounter clinical decision support tool. It ingests multi-source patient data, applies the Hyperspherical Patient Data Model, and surfaces conditions, risk scores, RAF uplift opportunities, and evidence-backed clinical actions for physician attestation — before the encounter begins.

WHAT DOES THE CLINICIAN DO?

Your role is to review, validate, and attest each EIE™-surfaced condition and clinical action. You are not approving an algorithm — you are applying your clinical judgment to AI-generated evidence. Your NPI-stamped attestation is the legal basis for documentation. The system supports; you decide.

REGULATORY CONTEXT

All attestations are governed by 42 CFR § 422.504(l) (MA physician attestation requirements), CMS RADV audit standards, and ICD-10-CM FY2026 Tabular coding guidelines. Declined conditions are logged for model calibration. All actions generate a 7-year audit trail.

⚡ STEP-BY-STEP ENCOUNTER WORKFLOW

STEP 1

Select Patient

Choose a preset profile or select Custom Patient to build from the 221-condition library. For real-world use, upload a patient record (HL7, CCDA, FHIR JSON, or CSV) via the Data Source Portal.

🏥 Patient Profile section

STEP 2

Review HPDM Scores

Examine the X/Y/Z axes and the Sphere visualization. X = condition breadth, Y = severity weight, Z = interaction multiplier (comorbidity compounding). The HPDM score vs. Legacy RAF delta reveals the documentation gap.

⚙ HPDM Engine · Sphere View

STEP 3

Triage Actions

Start with URGENT actions (24–72 hr window) before HIGH or STANDARD. Click the ⚡ EIE Brief badge to review net-new conditions the AI has surfaced that are not yet in the active problem list. Expand each action card for full evidence.

🔴 Action Priority Tiles

STEP 4

Attest & Cart

For each action, review the Financial, ICD-10, CPT, Quality, and References tabs. If the clinical evidence supports the condition, click Attest to add it to the 🛒 Cart. Decline or flag for dispute where appropriate.

✅ Attest → 🛒 Cart

STEP 5

Approve & Commit

Open the Final Encounter via the Cart. Review the full PCIB (Pre-Encounter Clinical Intelligence Brief). Add an optional clinician note, then submit your NPI-stamped approval. The completed encounter is printable and RADV-ready.

§4 Clinician Approval · NPI Stamp

C3S™ CONFIDENCE SCORE — READING YOUR ACTIONS

The Composite Condition Confidence Score (C3S™) reflects how strongly the multi-source evidence supports each condition. It is not a diagnostic score — it is an evidence-strength indicator to guide your review depth.

≥ 0.90HIGH — One-click attestation appropriate. Multi-source corroboration confirmed.

0.75–0.89MODERATE — Evidence review recommended. Check References tab before attesting.

0.65–0.74LOW-MOD — Active clinical confirmation required. Do not attest without chart review.

< 0.65LOW — Flag for investigation only. Do not attest. Consider a dispute or return.

ACTION PRIORITY TIERS — RESPONSE WINDOWS

🔴 URGENT 24–72 hrs

Immediate clinical intervention indicated. Includes overdose risk, hemodynamic compromise, critical safety-net flags (e.g., naloxone co-prescribing at MME ≥50), and acute decompensation patterns. Always address before HIGH or STANDARD items.

🟡 HIGH This visit

Significant RAF uplift and/or quality gap closure. Conditions with strong comorbidity interactions (Z-axis) or pending specificity upgrades (e.g., CKD Stage 3 → 3a/3b substage). Address within this encounter if clinically feasible.

🔵 STANDARD Next 30 days

Routine documentation opportunities, HEDIS/Star gap closures, CPT complexity uplifts, preventive screenings, and medication reconciliation items. Address in this or the next scheduled encounter.

ACTION CARD TABS — WHAT EACH CONTAINS

💰

FinancialRAF delta, PMPM impact, legacy vs. HPDM reimbursement gap, annualized revenue uplift

🏷

ICD-10Proposed ICD-10-CM codes, specificity upgrades, documentation requirements, RADV evidence

📋

CPTComplexity uplift codes, add-on codes, E&M level justification, complexity documentation cues

⭐

QualityHEDIS measure gaps, Star rating impact, VBP and ACO shared savings opportunities

📚

ReferencesPeer-reviewed clinical guidelines (AHA/ACC, KDIGO, USPSTF, ACS, AASLD) with DOI links

KEY TERMS

EIE™ — Encounter Intelligence Engine; AI layer that surfaces net-new conditions from federated sources

HPDM™ — Hyperspherical Patient Data Model; 3-axis risk scoring architecture (patent-pending)

RAF — Risk Adjustment Factor; CMS HCC v28 reimbursement coefficient

PCIB — Pre-Encounter Clinical Intelligence Brief; the EIE™ output delivered to the clinician

RADV — Risk Adjustment Data Validation; CMS audit framework requiring attestation evidence

Z-Axis — Comorbidity interaction multiplier; captures compounding risk from condition co-occurrence

⚠️ CRITICAL REMINDERS

🚫 Do not attest conditions you have not independently reviewed — the C3S score is a guide, not a clinical determination

📝 ICD-10 specificity matters — always select the most specific subcode supported by your documentation (e.g., N18.31 vs. N18.3)

⚖️ Disputes are tracked — if you believe a condition is incorrect or improperly coded, use the ⚖️ Disputes view; do not simply decline

🔒 EIE™ findings are locked evidence — original AI-surfaced evidence cannot be altered; your note is additive and stamped separately

🖨 Print or Save the Final Encounter — the .txt export is your RADV-ready documentation record; retain per your organization's 7-year policy

DATA SOURCE RELIABILITY HIERARCHY (Wᵢ weights used in C3S™ computation)

EHR Current 0.95 EHR Historical 0.92 Labs (LOINC) 0.91 Claims CMS 0.88 Pharmacy NCPDP 0.87 RPM Telemetry 0.85 NLP High-Conf 0.78 NLP Moderate 0.65 Free Text 0.55

Higher Wᵢ = less clinical scrutiny required · Lower Wᵢ = independent confirmation expected before attestation

Patient Profile

Margaret T.

Cardiorenal Diabetic

74y F

Robert J.

COPD + Pulmonary HTN

68y M

Dorothy M.

Metastatic Cancer

81y F

James W.

Cirrhosis + Coagulopathy

61y M

Sandra K.

Autoimmune + Renal

55y F

Thomas B.

Hypertension Only

67y M

Linda R.

Metabolic Triad

71y F

Carlos M.

Cardiopulmonary + BH

76y M

Custom Patient

Build Your Own

221 conditions

Source Instruments

HPDM™ Transformation Engine View: Sphere Axis Detail

Orbital View Condition Map ⚙ Engine View

X-AXIS

Breadth (conditions)

Y-AXIS

0.00

Severity weight

Z-AXIS

1.00×

Interaction multiplier

HPDM™ FORMULA

Score = Y-Axis Base × Π(1 + Z_i)

= 0.00 × 1.000 = 0.00

0.000

Legacy RAF

CMS HCC v28

0.000

Est. HPDM RAF

+$0/mo uplift

Confirmed Clinical Actions

URGENT

HIGH

STANDARD

⚡ EIE Brief

⚖️ Disputes

🛒 Cart

Indiana GROW · Hypersphere Core Platform

100K synthetic cohort · HRSA RFP July 2026 · Seed 42

Indiana GROW — 100,000 Patient Synthetic Cohort

Population calibrated to Indiana's rural Medicare/Medicaid profile using BRFSS 2023, CMS Geographic Variation PUF, and Indiana FSSA estimates. Seed 42 · March 2026.

Total Patients

100K

Synthetic cohort, seed 42

Conditions Modeled

221

Across 20 ICD-10 categories · HPDM v3 · 3 Cohorts Active · CT · IN · PA

Z-Axis Interactions

Comorbidity multiplier pairs

Action Rules

8 care categories · Tier A/B classified

Rural Patients

77.9%

RUCA-adjusted · IN baseline 72.4%

Dual-Eligible

38.2%

Medicare + Medicaid · IN baseline 35.1%

Mean Age

70.7

Skewed to 65–84 · aging rural population

Data Integrity

100%

Zero structural nulls · 45 fields complete

Complexity Distribution

Instrument	Model Type	Conditions	Z-Interaction	Mean Score	Corr w/ HPDM	Coverage Gap	Use Case
HPDM™	Multiplicative	221	65 Z-pairs	43.54	—	0	Pop health · RAF · Action
Charlson	Additive	19	None	1.76	0.356	202 conds	Mortality prediction
Elixhauser	Additive	31	None	4.25	0.309	190 conds	Hospital resource use
Epic	Additive	~80	Limited	14.40	0.434	~141 conds	EHR-embedded risk
Cerner	Additive	~75	Limited	14.26	0.418	~146 conds	EHR-embedded risk
IBM Watson	Additive	~100	Limited	13.54	0.421	~121 conds	Claims analytics

This cohort is specifically calibrated to Indiana's rural Medicare/Medicaid population using BRFSS 2023, CMS Geographic Variation Public Use File, Indiana FSSA data, and SHADAC 2024 estimates. Population parameters are designed to support the HRSA Indiana GROW RFP (July 2026 deadline).

Indiana GROW Program Context

The Indiana GROW (Growing Rural Outcomes with Wellness) initiative targets approximately 125,000 rural Medicare/Medicaid beneficiaries across 67 qualifying rural counties. The HRSA RFP requires submitters to demonstrate population health platform capability, risk stratification methodology, and care gap quantification at the county level.

This synthetic validation cohort represents 100,000 of the estimated 125,000 target population, calibrated to Indiana's specific disease burden, dual-eligibility rates, opioid crisis burden, and aging rural demographic profile.

IN vs. National

Indiana Prevalence Adjustments Applied

Condition	National Rate	Indiana Rate	HPDM Adjustment	Source	Indiana Impact
COPD	17.0%	21.8%	+28%	BRFSS 2023	Elevated CHF+COPD Z-axis firing
Opioid Use Disorder	2.1%	3.5%	+65%	CMS State Drug Data	PTSD+Opioid Z-axis elevated
Diabetes (Complicated)	9.4%	11.1%	+18%	IN Dept of Health	DM+CKD interaction more prevalent
Depression/MDD	8.3%	10.1%	+22%	BRFSS 2023	Rural mental health · Dementia+Depression
Hypertension	47.4%	53.1%	+12%	CMS Geo Var PUF	HTN+CKD interaction elevated
Obesity (BMI≥30)	30.7%	36.8%	+20%	BRFSS 2023	NAFLD+DM Z-axis trigger elevated
Tobacco Use Disorder	12.5%	19.5%	+35%	IN Dept of Health	COPD severity amplification
PVD / PAD	4.5%	5.2%	+15%	SHADAC 2024	Rural vascular disease burden

RFP Deadline

July 2026

HRSA Indiana GROW
~$125M federal funding

Federal Stakeholders

HRSA + CMS

CMS Quality Demonstration
HRSA Rural Health Grant

Lead Entity

C4E

Care4Everyone as proposal lead
Hypersphere for HPDM/data only

HPDM™ PLATFORM — EXECUTIVE BRIEFING

Hypersphere Health Population Data Model
Indiana GROW Validation Report

HPDM™ is a multiplicative, three-axis population health scoring model federating data across 221 clinical conditions, 65 Z-axis comorbidity interactions, and 27 care action rules from 5 source instruments. The Indiana GROW synthetic cohort of 100,000 patients demonstrates a $310.2M annual reimbursement gap and 108.5% mean Z-axis risk lift vs. legacy additive instruments.

Patent-Pending v13 ICD-10-CM 2024 CMS HCC v28 2025 FHIR R4 Aligned Phase 1 Validation Complete

Annual Reimb Gap

$310.2M

vs. legacy instruments

Conditions Federated

221

20 ICD-10 categories

Z-Axis Interactions

0 in any legacy system

Patients w/ Uplift

88.1%

+22% RAF mean

🔬

Clinical Depth

▸ 221 conditions across 20 ICD-10 categories
▸ 65 comorbidity multiplier pairs (Z-axis)
▸ 5 source instruments federated in real-time
▸ LOINC · SNOMED · RxNorm · FHIR R4 aligned

💰

Financial Intelligence

▸ $310.2M annual gap across 100K cohort
▸ CMS MA v28 RAF rate book integration
▸ PMPM savings from CMMI 2022 demonstration
▸ 27 billable action rules with CPT + ICD-10

⚡

Operational Scalability

▸ Real-time IRIS for Health federated query
▸ Multi-region YAML replication (~8 min/region)
▸ SOC 2 Type II roadmap (Phase 2)
▸ HITRUST + ONC certification in plan

🗺️ Next Steps — Build to Scale Roadmap

Every step from current state to enterprise scale. In order. Costs estimated. Sources linked. No step is skipped — each unlocks the next. Click any section to expand.

● Phase 0 — Desktop Demo · Right Now $0 DO THIS FIRST

▼

Install Docker Desktop

Unlocks full v13 shell — all 7 engines, no 6-tab ceiling, Delta Analytics live. Most important action. Free. 15 minutes. Allocate 8GB RAM in Docker → Settings → Resources.

📥 docker.com/products/docker-desktop

Free

Run Platform Shell

After Docker installs: Terminal → navigate to Hypersphere folder → one command. Platform at localhost:3000 with all engines. Confirm all 7 engines render.

💻 Command: docker compose up -d → http://localhost:3000

Free

GitHub Private Repository

Version control for every session. Upload Hypersphere_Repository.zip contents to private repo. Add session files to Claude Project for persistent context.

📥 github.com → New Repository → Private → Upload files

Free

GROW RFP Evidence Review

Indiana GROW HRSA RFP due July 2026. Review synthetic 100K cohort evidence in Geography Explorer + Delta Analytics. Confirm completeness for RFP technical evidence package.

📋 hrsa.gov/grants/find-funding · HRSA GROW program page · RFP deadline: July 2026

Free

◎ Phase 1A — Shared Demo + Back-End $18–80/mo Next

▼

hypersphere.health or demo.hypersphere.health. Cloudflare Registrar cheapest — includes free DNS + DDoS. Professional URL for health plan demos.

📥 cloudflare.com/products/registrar · ~$12/year

~$12/yr

DigitalOcean VPS — First Shareable URL

2 vCPU, 4GB. Same Docker stack, now accessible 24/7. Partners click link without you present. Sign BAA before any data touches server.

📥 digitalocean.com · $18/mo · BAA: Settings → Legal

$18/mo

FastAPI Back-End Stub

Python FastAPI: POST /score, /delta/run, /apcd/generate returning mock data. Front-end connects to real APIs. Claude builds this with you in one session.

📦 fastapi.tiangolo.com · Free · 1 session with Claude to build

Free

PostgreSQL + Synthetic Data Load

Create hpdm schema: patients, score_archive, claims_outcomes, apcd_state_config. Load 100K synthetic Indiana patients. HPDM scoring runs. Delta validation live.

📦 postgres:16 Docker image · Schema in backend/README.md · Claude builds init script

Free

Redox Sandbox Account

Free sandbox tests payer/EHR connectivity. Validates integration architecture before going live. Start now to understand 60–90 day credentialing timeline.

📥 redoxengine.com → Contact Sales → sandbox · Free sandbox · Live feeds: $500–2K/mo

Free sandbox

Indiana FSSA Data Access Agreement

Required for real Indiana cohort data for GROW RFP. 60–90 day approval. Start immediately for July 2026 deadline.

📥 in.gov/fssa → Data request form · 60–90 day timeline · Start NOW

Free (application)

◎ Phase 1B — IRIS + First Contract $800–2,500/mo

▼

InterSystems IRIS for Health — Request Pricing

Submit questionnaire: transformation + native multi-model clinical platform. Request BYOL EC2 r6i pricing. 4 vCPU Phase 1B, 16 vCPU Phase 2 HA. Request FHIR SQL Builder technical demo. Sales process — start early.

📥 intersystems.com/iris-for-health → Contact Sales · BYOL ~$3K–8K/mo

$3K–8K/mo

AWS EC2 + BAA — HIPAA-Eligible Infrastructure

EC2 r6i.xlarge (4 vCPU, 32GB). Sign AWS BAA. Enable CloudTrail. RDS PostgreSQL for config store. First infrastructure that can touch real PHI.

📥 aws.amazon.com/compliance/hipaa-compliance · EC2 ~$180/mo · RDS ~$150/mo

~$350/mo

Sutter Health Engagement

$4.3M–$7.9M Phase 1 value proof-point. EHR via Redox (Epic). Data sharing agreement required. Use HPDM Delta Analytics latent risk evidence as conversation opener.

🤝 Proof-point: 26.4 composite → Imminent → 68% hosp rate predicted. Evidence in Delta Analytics tab

Relationship

⚠️ Northwood Health — Confirm Entity Details

CLARIFICATION NEEDED: What is Northwood Health? Health system, hospital, IPA, or ACO? What market? What EHR? Approximate patient population? Answer determines engagement strategy and full specification.

📋 Action: Confirm entity type → Claude builds Northwood-specific reporting engine and cohort tile

Spec Needed

NCQA HEDIS License

Required for commercial HEDIS reporting to any health plan or health system under contract. Annual license. Apply 90 days before first health plan contract closes.

📥 ncqa.org/hedis → Licensing · $5K–50K/yr depending on org size

$5K–50K/yr

⬡ Phase 2 — Enterprise Scale $5K–15K/mo After First Contract

▼

CareFirst BCBS — Warm Sponsor Required

Win probability: 35–50% with warm sponsor; 15–25% cold. Priority: identify internal CareFirst champion. 3 work streams: quality/performance, MA growth/D-SNP, enterprise governance.

🤝 CareFirst HQ: Baltimore, MD · Contact: VP Population Health or Medical Director · BAA required before data

Sponsor Required

SCAN Health Plan Engagement

West Coast MA. CA/AZ/NV/TX. Known for innovation. HPDM latent risk directly applicable to MA HCC capture revenue. Value prop: HCC uplift + Stars via risk identification.

🤝 SCAN HQ: Long Beach, CA · Contact: CMO or VP Population Health

Relationship Required

IRIS HA Cluster + Begin HITRUST

After first large contract: IRIS 16 vCPU HA cluster. Sub-second at 1M+ patients. Begin HITRUST CSF (12–18 months). Required for enterprise health plan procurement.

📥 hitrustalliance.net · 12–18 months · ~$50–100K · Start at $500K+ contract

$50–100K one-time

SOC 2 Type II — Run Parallel with HITRUST

Required by most enterprise health plan procurement. 6–12 months. Run parallel with HITRUST. Both certifications together satisfy virtually all enterprise security requirements.

📥 vanta.com (automated prep) · ~$10–30K/yr · Start after Series A or $1M ARR

$10–30K/yr

Technical Co-Founder / Fractional CTO

The one human gap Claude cannot fill: someone who deploys code, manages infrastructure, interfaces with vendors. 20 hrs/month until Phase 2. Claude + this person = full engineering stack.

🤝 Profile: Python/FastAPI + Docker/AWS + healthcare data (FHIR, HL7v2) · Platforms: Toptal, Gun.io · Rate: $100–150/hr

$2K–3K/mo

Active Program: CT RHT · Care 4 Connecticuters · PC-CT-01 · v1 Active

Program Population

323,304

rural residents

Managed MCC Pts

80,826

in-scope MCC

RAF Delta

0.238

legacy vs HPDM

Annual RAF Gap

$265.5M

revenue opportunity

V28 Adjusted Gap

$257.2M

CMS-HCC V28 2026

5-Year Value

~$1.3B

RAF + clinical + RPM

HPDM Risk Tier Distribution · PC-CT-01

Imminent

9.0%

7,274

Critical

19.0%

15,357

High

22.0%

17,782

Moderate

26.0%

21,015

Stable

24.0%

19,398

Baseline vs Ecosystem-Enabled Trajectory

Baseline Population Trajectory

Ecosystem-Enabled

Cross-Cohort Comparison — CT · IN · PA Active Program Cohorts

No circular validation — cross-cohort references confirm formula consistency only. Each cohort independently anchored to CMS-HCC V28 2026 and CMS PUF data.

METRIC	CT RHT	IN GROW	PA RHT	HPDM CROSS-COHORT REF
Program Population	323,304	500,000	2,400,000	—
Managed MCC Patients	80,826	100,000	1,488,000	—
RAF Delta (Legacy → HPDM)	0.238	0.225	0.230	0.231 ± 0.028
HPDM RAF Mean	1.335	1.445	1.410	1.397 ± 0.056
PMPM Delta (Leg → HPDM)	$273	$259	$265	$266 ± $7
Annual RAF Gap	$265.5M	$310.2M	$1.935B	$2.511B total
V28 Adjusted Gap (2026)	$257.2M	$300.6M	$1.875B	$2.432B total
Attainment Rate (HPDM)	88.1%	88.1%	88.1%	88.1% (formula consistent)
RVC Assignment	RVC-01 Northeast	RVC-04 Great Lakes	RVC-02 Mid-Atlantic	3 RVCs active

CMS-HCC V28 2026 Adjustments (operative): CKD 3A −50.4% · COPD −18.8% · Morbid Obesity −19.4% · These reductions amplify the HPDM gap identification advantage — conditions removed from legacy RAF are captured in HPDM Z-axis interactions.

Existing Geographies

Indiana GROW

100,000 pts · $310.2M gap

Active

Michigan Rural

82.1% rural · Template

Template

Connecticut C4C

323K pop · $125M · C4E

Template

California FQHC

24.6% rural · Template

Template

🔬 PATIENT ENGINE DRILL-DOWN

Load a representative individual from the active geography into the HPDM™ Patient Engine. Demonstrates full 221-condition × 3D-axis computation at the individual level.

Margaret T.

Cardiorenal Diabetic Complex · Indiana GROW

64.13

HPDM Score · Critical

CHF CKD-Moderate DM-Complex HTN Anemia Depression

Conditions

Z-Pairs

1.357

Legacy RAF

LOAD MODE

🌐

Geo Sample

Mode A

📥

Upload Record

Mode B

🔧

Build Profile

Mode C

Opens Patient Engine · Engine View pre-activated · Geography context preserved

Geography Type

🗺️
County / Rural

🏥
FQHC / Health Center

💳
Payer / Health Plan

🩺
Provider / IPA

🏛️
HHS Region

📍
State / Multi-State

Input Parameters

GEOGRAPHY NAME

STATE

POPULATION SIZE

MEAN AGE

RURAL % (RUCA)

DUAL-ELIGIBLE %

ANALYST NOTES

📡 HPDM™ Geography Data Sources

Available sources increase confidence · Unavailable sources flagged as HPDM standard gaps

HPDM CONFIDENCE SCORE 0%

No supplemental data connected — parametric estimate only

✅ CORE — AVAILABLE TODAY

🔶 STANDARD AUGMENTATION — Optional Upload

🔮 ADVANCED — HPDM STANDARD (Emerging / Future)

Not yet broadly available · Connection unlocks full HPDM precision · Currently flagged as confidence gaps

📁 Primary dataset

CMS PUF · BRFSS · HCUP NIS

📁 Supplemental dataset

SDoH · RPM · Claims extract

HPDM STANDARD GAPS — Connect to maximize confidence

UPLOAD GEOGRAPHY DATA FILE

📤

Click to upload geography dataset

Core: CMS PUF · BRFSS · HCUP · ACS · RUCA
Standard: Claims · EHR · Pharmacy · APCD · HEDIS
Advanced: RPM · SDoH · CCM · TEFCA

Geography Data Input

Upload cohort data or select preset geography to generate all views

HPDM v3 · 3 Cohorts Active · CT · IN · PA · 221 Conditions · 20 Categories

📥

Upload Cohort Data

CSV or JSON with patient-level or aggregate data. HPDM will map fields and identify gaps in the 221-condition taxonomy.

✓ CSV · JSON · HL7 FHIR Bundle

🗺️

Use Available Geography

Select from pre-loaded geographic cohorts. All views auto-populate with validated synthetic data for the selected region.

✓ Indiana GROW · Michigan · Connecticut · California

HPDM Taxonomy Completeness

221 conditions · 20 ICD-10 categories

Identifies missing data across HPDM's 221-condition taxonomy — mirroring the 21-chapter ICD-10-CM architecture while excluding 70,000+ administrative billing codes.

🌐

Configure a Geography and Run Analysis

Enter population parameters for any U.S. geography — county, FQHC, payer, HHS region, or state. HPDM will generate a calibrated validation profile using the full 221-condition library and 65 Z-axis interactions.

⚡

EIE™ ENCOUNTER INTELLIGENCE ENGINE — HPDM™ Patent-Pending

Loading geography...

Open the Patient Engine tab to interact with EIE™ in real time.

C3S™ Composite Condition Confidence Score

C3S = [Σ(Wᵢ × Eᵢ × Lᵢ)] / Σ(Wᵢ) × Kcorr

Wᵢ Source reliability weight0.55–0.95

Eᵢ Extraction confidence0.65–1.00

Lᵢ Longitudinal factor1.00–1.40

K_corr Corroboration multiplier1.00–1.28

Confidence Thresholds

≥ 0.90 HIGH — one-click attestation

0.75–0.89 MODERATE — evidence review

0.65–0.74 LOW-MOD — active confirmation

< 0.65 LOW — investigate only

Source Reliability Weights (Wᵢ)

Tier 1 — Algorithmic

· C3S confidence gating

· Contradiction detection

· Temporal validity check

· Med-condition consistency

Tier 2 — Physician Attestation

· MD/DO sign-off (42 CFR § 422.504(l))

· Decline tracking → model calibration

· Same-level clinician amendments

· NPI-stamped encounter commit

Tier 3 — Audit Trail

· Full provenance chain per condition

· RADV-ready documentation

· 7-year record retention

· Amendment log + regulatory window

v13 — DATA INTEGRATION ARCHITECTURE · Alpha Omega Health & AkēLex CDS

HPDM™ EIE™ FEDERATION PIPELINE — v13 Partner Integration Points

SOURCE LAYER

📡 Alpha Omega Health

RPM Device Telemetry
Aware Inc. Biometrics
Edge Processing Layer

LIVE · Wᵢ 0.85

🏥 EHR / Claims

Epic · Cerner · Elixhauser
CMS Part A/B Claims
LOINC Lab Results

Wᵢ 0.88–0.95

💊 Pharmacy / SureScripts

NCPDP Fill History
MME Calculations
Formulary Data

Wᵢ 0.87

→

CDS ENRICHMENT LAYER

⚖️

AkēLex CDS Layer

SIMULATED · Architecture Placeholder

PENDING API

Injects into EIE pipeline:

• Contraindication hard-stop rules (NSAID/CKD, Metformin eGFR)

• HEDIS gap trigger logic (A1c, eye exam, statin, BP control)

• MME auto-calculation + naloxone co-prescribing triggers

• ICD-10 specificity upgrade recommendations

• Formulary alignment + drug-level monitoring triggers

• Anticoagulation CHA₂DS₂-VASc + HAS-BLED auto-scoring

Connection Point

HPDM EIE™ ← AkēLex REST API
Credential exchange: pending
Integration pathway: FHIR CDS Hooks

→

HPDM™ EIE™ CORE

⚡ Encounter Intelligence Engine

• C3S™ confidence scoring (Wᵢ × Eᵢ × Lᵢ × Kcorr)

• HPDM 3-axis risk scoring (X/Y/Z)

• Z-axis comorbidity interaction matrix (65 pairs)

• 221-condition library · HCC v28 RAF mapping

• 27 clinical action rules across 8 care categories

• Protocol derivation → Care Team routing

Outputs:

• Pre-Encounter PCIB (clinician-facing)

• Confirmed Clinical Actions (27 rules)

• Care Captain™ dispatch payload

• RADV audit package

→

CARE TEAM LAYER

🧭 Cromford Care Captain™

Cross-team dispatch hub
Task routing + tracking
Simulated in v13

👨‍⚕️ MD / NP / RN

Clinical actions
Orders · Attestation
Rx · Protocols

🤝 MSW · OT · PT · RD

Allied health referrals
SDoH interventions
Functional assessments

📋 PharmD · CDCES · CPC

Pharmacy · Education
Coding & RADV
Non-clinical support

■ Alpha Omega — LIVE integration · FHIR R4 Push · Wᵢ 0.85 ⊡ AkēLex — Simulated input · FHIR CDS Hooks pathway · API handshake pending ◎ Care Captain™ — Simulated dispatch · Cromford Health · v13 demo Hypersphere Core Platform v13 · Patent-Pending · Hypersphere Health™ · Naya Advisory Services

📡

Alpha Omega Health — RPM Integration

Live data pathway · Aware Inc. biometrics · Wᵢ = 0.85 in C3S™ computation

ACTIVE

Device Data

· Daily weight
· BP (systolic/diastolic)
· O₂ saturation
· Heart rate / rhythm
· Temperature
· Activity (accelerometer)
· Glucose (CGM integration)
· Respiratory rate

Integration Path

· Alpha Omega edge device → FHIR R4
· Push to HPDM EIE™ federation layer
· C3S Wᵢ 0.85 applied at ingestion
· Aware Inc. biometric processing
· HPDM alert threshold monitoring
· Post-dispatch: RPM protocol activation
· Patient-facing: Care Captain alerts
· 7-year audit trail per encounter

Alpha Omega Health · Aware Inc. biometrics · HPDM EIE™ — protocol-specific monitoring activated per Care Captain dispatch

⚖️

AkēLex — CDS Integration Placeholder

Simulated input · Architecture defined · API handshake pending

PENDING

CDS Inputs (Simulated)

· Contraindication hard-stops
· HEDIS gap trigger rules
· MME + naloxone triggers
· ICD-10 specificity guidance
· Drug-level monitoring alerts
· Formulary alignment rules
· CHA₂DS₂-VASc + HAS-BLED
· Beers Criteria polypharmacy

Integration Architecture

· Protocol: FHIR CDS Hooks R4
· Trigger: EIE™ condition confirmation
· Endpoint: AkēLex REST API (TBD)
· Auth: OAuth 2.0 client credentials
· Response: JSON CDS card payload
· EIE injects card into action detail
· Credential exchange: in scoping
· Go-live: pending partner agreement

AkēLex · CT Clinical Decision Support · Integration pathway defined in HPDM v13 — live inputs simulated for Indiana GROW HRSA RFP demonstration

Core Platform Architecture v13

Open-Source FHIR R4 Stack · HAPI FHIR + PostgreSQL · Redox + Airflow · Docker-Native · Phased Build Strategy · v13

● IRIS REMOVED ● LEADINGDOTS REMOVED HAPI FHIR NATIVE DOCKER CONTAINERIZED

View by Phase:

All Phases — Architecture Comparison

Full stack comparison across all three data layer phases. Click a phase button above or a phase card below to view the full architecture diagram.

● Phase 1A · Now

Open-source AWS stack. HAPI FHIR + PostgreSQL. Batch scoring via Airflow. Running today on Docker. Adequate for 120-day contract and GROW RFP.

Stages 1–2 · View Architecture →

◎ Phase 1B · IRIS Migration

IRIS for Health replaces HAPI FHIR + PostgreSQL. FHIR SQL Builder enables zero-copy live scoring. Triggered at client go-live milestone.

Stage 3 · View Architecture →

⬡ Phase 2 · Full Scale

IRIS HA cluster. Sub-second at 1M+ patients. HITRUST/SOC 2. Full APCD delta on FHIR SQL. Enterprise compliance.

Stage 4 · View Architecture →

Platform Layer

Phase 1A · Now

Phase 1B · IRIS

Phase 2 · Scale

Ingest

Redox Engine · FHIR R4 normalization

Redox + Health Connect · DTL/BPL inside IRIS

Health Connect (scaled) · Redox for payer feeds

FHIR Storage

HAPI FHIR Server · Open-source · Docker

IRIS for Health · Native FHIR R4

IRIS for Health · HA Cluster

Clinical Storage

PostgreSQL 16 · Score Archive · Claims

IRIS Globals · Native sparse multi-dimensional

IRIS Globals · HA · Point-in-time recovery

Analytics

Airflow batch · DuckDB → PostgreSQL

FHIR SQL Builder · Zero-copy · No ETL

FHIR SQL Builder + APCD delta · Multi-tenant

Scoring

Batch-approximated · 5–15 min latency

Near real-time · Sub-second · Live queries

Sub-second · 1M+ patients · Multi-tenant

Latency

5–15 minutes

< 1 second

< 1 second at scale

Infrastructure

Docker · $0–80/mo · Synthetic data

AWS EC2 r6i · IRIS BYOL · HIPAA eligible

EC2 HA or Azure VM · Encrypted · Audit logs

Compliance

None (synthetic) · BAA at Pilot

HIPAA eligible · BAA active

HITRUST CSF · SOC 2 Type II · Full ITSM

Est. Cost

$0–80 / month

$800–2,500 / month

$5,000+ / month

App Code Changes

Baseline

Zero · FHIR R4 REST + JDBC unchanged

Zero · Only base URL env var changes

▲ v13 UPDATES IRIS → HAPI FHIR (open-source) · Leadingdots → Redox + Airflow · Added: APCD Module · Delta Analytics · Score Archive · FastAPI · Infra: Docker · Desktop→Enterprise path · Architecture: Shell-and-iframe · ADR Phased Build Strategy Platform v13

⬡ ADR-001 · Architecture Decision Record — Data Layer Selection: Phased Build Strategy

Hypersphere Health, Inc. · March 2026 · Internal — Architecture Decision Record · Status: DECIDED

● DECIDED PHASE 1A · ACTIVE

Decision

Adopt a phased build strategy. Begin with an open-source AWS stack for initial development and contract execution. Migrate to an InterSystems IRIS for Health data layer post-contract and pre-launch, when live scoring capability is required. The application layer — scoring engine, APIs, care management interface — is fully portable across both data layer implementations.

Key Constraint

The Hypersphere scoring engine requires live queries against longitudinal patient data — clinical, claims, and RPM — with sub-second response times. This is not achievable with ETL-dependent architectures at 250K–1M patient scale. The current AWS stack introduces 5–15 minute minimum latency between a clinical event and scoring engine awareness.

Phased Build Strategy

Phase 1A — AWS Stack ● NOW ACTIVE

IngestHealth Connect

StorageHAPI FHIR + PostgreSQL

AnalyticsBatch pipeline (Airflow)

ScoringBatch-approximated

MPIAWS Entity Resolution

Latency5–15 min (ETL lag)

Adequate for 120-day contract window. Batch scoring acceptable for GROW RFP submission.

Phase 1B — IRIS Migration ◎ PRE-LAUNCH

IngestHealth Connect (unchanged)

StorageIRIS for Health

AnalyticsFHIR SQL Builder — zero-copy

ScoringNear real-time live queries

MPIAWS Entity Resolution

LatencySub-second (no ETL)

Migration trigger: go-live milestone requiring live scoring. App code does not change — IRIS exposes standard FHIR R4 REST + JDBC/ODBC.

Phase 2 — Full Scale ⬡ FUTURE

IngestHealth Connect (scaled throughput)

StorageIRIS for Health — HA cluster

AnalyticsFHIR SQL Builder + APCD delta

ScoringSub-second, multi-tenant

MPIAWS Entity Resolution

LatencySub-second at 1M+ patients

Portability guarantee: app code unchanged. Only base URL (env variable) changes at each migration stage.

Why the AWS Stack Cannot Support Live Scoring

Failure Mode	AWS Stack (Phase 1A)	IRIS Stack (Phase 1B+)
ETL Pipeline	2-hop ETL: HealthLake → PostgreSQL → DuckDB → scoring engine	Zero-copy — scoring engine queries live FHIR directly via SQL
Latency	5–15 min minimum between clinical event and score awareness	Sub-second — no pipeline, no lag
Silent failure	Pipeline failures produce stale scores without alerting care team	Direct query — failure is immediate and visible
Schema rigidity	PostgreSQL requires schema engineering for sparse hierarchical data	IRIS Globals — native sparse storage, no padding required
FHIR SQL	✗ Not available — no direct SQL against live FHIR resources	✓ FHIR SQL Builder — direct SQL against live FHIR R4
Multi-model	✗ Claims, clinical, RPM across 3 separate systems	✓ Single engine — HL7v2, FHIR, claims, RPM in one store
Scale ceiling	✗ Sub-second scoring at 1M patients not achievable	✓ Sub-second at 1M+ patients — core requirement met

Risk Register

Risk	Level	Description	Mitigation
Vendor lock-in	HIGH	IRIS becomes critical path at Phase 1B	Maintain portable app layer; FHIR R4 REST interfaces are standard; data export to FHIR R4 always possible
License cost	MEDIUM	IRIS for Health license > Health Connect pricing	BYOL on EC2 vs. SaaS; HealthLake cost eliminated at Phase 1B partially offsets delta
Migration timing	MEDIUM	Phase 1A → 1B window may compress under contract pressure	Phase 1A app development accelerates timeline; IRIS deployment is parallel workstream
Phase 1A scoring gap	LOW	Batch scoring insufficient for some live use cases	GROW RFP timeline: Phase 1A acceptable for initial submission; live scoring required at client go-live

Self-Build Assessment — ~70% Hypersphere-Owned Regardless of Data Layer

Component	Primary Tooling	Portability
Scoring Engine	Python / FastAPI	Queries IRIS via FHIR SQL — no change at Phase 1B
FHIR R4 API Layer	FastAPI + IRIS FHIR REST	Standard FHIR endpoints — fully portable
Care Management Interface	React + REST calls	No data layer dependency
APCD Delta Analytics	Python on IRIS FHIR SQL	Replaces DuckDB ETL — simpler at Phase 1B
Reporting Dashboards	React / Recharts / D3	Fully portable — no data layer dependency
Pipeline Orchestration	Python / Airflow	Simplified at Phase 1B — fewer ETL jobs required
API Gateway & Auth	FastAPI / OAuth2 / SMART	Unchanged — tenant context in JWT
InterSystems-owned (not self-buildable)	ObjectScript / DTL / BPL transforms · IRIS namespace config · DB engine · HA cluster tuning — licensed, not built (~15–20% of platform footprint)

Immediate Next Steps

Submit InterSystems questionnaireScoped to transformation + native multi-model clinical data platform capabilities

Request IRIS for Health pricingBYOL on EC2 r6i — 4 vCPU Phase 1B, 16 vCPU Phase 2 HA cluster

Continue Phase 1A self-buildHAPI FHIR + PostgreSQL open-source stack — in progress, v13 delivered

Define Phase 1B migration triggerIdentify the specific go-live milestone that requires live scoring to activate migration

Validate FHIR SQL BuilderRequest InterSystems technical demo against multi-tenant namespace with live FHIR R4 data

📥 Zone 1 — Data Sources All inbound streams normalize to FHIR R4 before entering the platform

EHR

EHR Systems

Epic · Cerner · athenahealth · Any FHIR

Primary clinical data source. FHIR R4 REST API primary; HL7 v2 for legacy. Redox normalizes all to FHIR before platform ingestion.

FHIR R4HL7 v2Via Redox

Primary Source

EoB

Medical Claims

FHIR ExplanationOfBenefit (Medical)

837P/I claims normalized to FHIR R4 EoB. ICD-10 diagnoses, CPT/HCPCS procedures, NPI, financials. Ingested via fhir_eob_parser.py.

FHIR R4 EoB837P/INEW

Phase 2 Added

Pharmacy Claims

FHIR EoB (Pharmacy) / NCPDP D.0

NDC codes, days supply, ingredient cost, dispensing fee, plan paid. NCPDP D.0 normalized to FHIR EoB pharmacy profile at ingestion.

FHIR R4 EoBNCPDPNEW

Phase 2 Added

834

Eligibility / Enrollment

FHIR Coverage + Patient / 834

Member demographics, plan enrollment dates, coverage type, dual-eligible flag. 834 normalized to FHIR R4 Coverage + Patient resources.

FHIR Coverage834NEW

Phase 2 Added

NPI

Provider Master

FHIR Practitioner / Organization

NPI registry, taxonomy codes, network status, rural designation, ACO participation. FHIR R4 Practitioner and PractitionerRole resources.

FHIR R4NPI RegistryNEW

Phase 2 Added

RPM

RPM Device Telemetry

Alpha Omega IoT365 · Aware Biometrics

Daily weight, BP, O₂ saturation, glucose, activity. FHIR Observation push. C3S™ Wᵢ 0.85. ACTIVE integration.

ACTIVEFHIR R4IoT

Live

↓ All sources → Redox Engine (payer connectivity) → normalized to FHIR R4 → HAPI FHIR Server

⚙️ Zone 2 — Ingestion Pipeline Phase 1A · HAPI FHIR + PostgreSQLTransforms at Phase 1B → IRIS for Health

RDX

Redox Engine

Payer + EHR Connectivity Partner

Credentialed payer connections for 837/834/NCPDP feeds. Normalizes to FHIR R4 before handing off. Eliminates the hardest data acquisition problem.

FHIR R4HIPAA12,200+ Orgs

PartnerReplaces Leadingdots ETL

H⊕

HAPI FHIR Server

Open-Source · Self-Hosted · Docker

Full FHIR R4 compliant server. Replaces InterSystems IRIS for Health. Runs in Docker container. Native EoB, Coverage, Practitioner resource support. Apache 2.0 license.

FHIR R4Open-SourceReplaces IRIS

Core Platform

PostgreSQL 16

Primary Data Store · Open-Source

HPDM Score Archive, claims tables, members, enrollments, APCD config, ingestion logs. Migrates to AWS RDS or Azure Database with zero code changes.

PostgreSQLOpen-SourceReplaces IRIS DB

Core Platform

Apache Airflow

Pipeline Orchestration · Open-Source

5 DAGs: daily EoB ingestion, enrollment sync, nightly HPDM scoring, weekly delta computation, monthly APCD generation. Replaces Leadingdots-managed pipelines.

Open-Source5 DAGsNEW

Phase 2 Added

🐳

Docker Compose

Full Stack Containerization

Entire platform starts with one command. Same containers deploy from laptop to VPS to enterprise cloud. OCI standard — runs on any infrastructure.

Open-SourceOCI StandardDesktop Ready

Infrastructure

↓ Normalized FHIR R4 longitudinal record stored in HAPI FHIR + PostgreSQL → Airflow batch pipeline triggers HPDM scoring

🔮 Zone 3 — HPDM Engine — Hypersphere Proprietary IP · Patent-Pending · Python 3.11 The only component not replicable through third-party licensing

Hyperspherical Patient Data Model

3-Axis Compounding Intelligence · 221 Conditions · 205 SDoH Factors · 510 Real-Time KPIs · 59 Clinical Instruments

X — Axis

Classification

Condition subtype, drug class, SDoH domain, test type. Not just "diabetes" — "Type 2 DM with end-organ damage, A1C 9.4%, declining trend."

Y — Axis

Prevalence

Epidemiological context — population prevalence, community exposure, adherence benchmarks. Rural ≠ Urban. Geography changes the risk picture.

Z — Axis

Severity / Compounding

Clinically validated severity PLUS multiplicative interaction effects across all active conditions, SDoH, medications, and engagement. 59 instruments · 203 severity tiers · 177 compounding examples.

221 Conditions205 SDoH Factors510 KPIs59 Instruments5 Risk Tiers65 Z-InteractionsRAF/HCC Mapped100K Indiana ValidationAPCD-ReadyDelta Analytics Engine

↓ HPDM computation → Score Archive (immutable) + 12 FHIR R4 output types → Delta Engine + APCD Module

🆕 Zone 4 — New Platform Components (Phase 2 Additions) Self-validating feedback loop · APCD compliance · Evidence generation

🗄️

HPDM Score Archive

Immutable · Timestamped · PostgreSQL

Every HPDM computation archived at the moment of scoring — before any claims outcome is known. Non-negotiable foundation for Delta Analytics validation integrity.

Fields: composite score, z/x/y axes, active domains, instrument scores, interaction weights, engine version
Policy: Immutable — corrections create new records with supersession pointer
Retention: Indefinite — this is the historical validation dataset

PostgreSQLImmutableNEW

Phase 2 Added

📋

APCD Submission Module

State-Configurable · JSON-Driven

Generates state-spec-compliant APCD flat files from HPDM data store. Adding a state is a JSON config record — zero code changes.

Files: Medical claims, pharmacy claims, eligibility, provider master, manifest
States configured: Indiana (IDOH) · Connecticut (OHS)
Validation: Referential integrity, NPI format, ICD-10 format, amount sign checks

IndianaConnecticutNEW

Phase 2 Added

Delta Analytics Engine

Prediction vs. Reality · Self-Validating

Joins Score Archive predictions against realized claims outcomes. Proves HPDM predicts hospitalization before claims show it. The evidence-generation engine.

Outputs: Tier accuracy, cost validation, latent risk patients, ROI quantification, weight calibration signals
Validation: 25.8× cost separation (Imminent vs Stable), all 5 tiers calibrated
ROI: $2.05M net benefit at 35% efficacy — Imminent + Critical tiers

Python/DuckDBValidatedNEW

Phase 2 Added

API

FastAPI REST Layer

20 Endpoints · OpenAPI 3.0

Lightweight REST API exposing HPDM engine, FHIR store, APCD formatter, and Delta Engine to the React frontend and partner integrations.

Endpoints: POST /score, GET /score/{id}, POST /apcd/generate, POST /delta/run, GET /delta/summary
Docs: Auto-generated OpenAPI at /docs

FastAPIOpenAPINEW

Phase 2 Added

↓ HPDM outputs consumed by plug-in components · All plug-ins connect via FHIR R4, HL7 v2, or REST API

🔌 Zone 5 — Plug-In Component Framework Optional · No plug-in is required · All connect via FHIR R4 or REST

Care Management Operations

Alpha Omega Health

Native: HPDM outputs available via API to any care management system

Hx360 CCM/RPM billing · IoT365 device telemetry writes to FHIR · 24/7 virtual care. ACTIVE integration at Wᵢ = 0.85.

ACTIVEFHIR R4RPM/RTM

Virtual Care Delivery

Cromford Health

Native: HPDM outputs routed to any virtual care platform

Care Captain™ dispatch hub · ICT workflow · HealthPursuits™ · 23+ countries · d.b.a. Telemedicine.com

NationalVirtual ICURTM/CCM

Care Pathway Engine

Individuallytics

Native: HPDM ACIs routed to care team via API output

N-of-1 pathway micro-adjustment · HealthPursuits™ 2-week sub-goal generation · Patient longitudinal data

Patent-PendingN-of-1

CDS Layer (CT-Specific)

AkēLex

Native: ACIs formatted for physician consumption via API

Role-aware real-time CDS at point of care · FHIR CDS Hooks R4 · CT/ME markets · Integration pending

PENDINGCDS Hooks

Patient Engagement (CT-Specific)

Prosumer Health

Native: Patient-facing FHIR R4 API for any patient app

Human-guided health coaching · AI symptom triage · Patient-reported SDoH Z-code collection · CT/ME

CT/MEPatient App

⚠ Reduced Dependency

InterSystems IRIS

Now optional — only where state HIE already runs IRIS

If client/state HIE already runs InterSystems, IRIS feeds HAPI FHIR as an inbound data pipe — not the store. Core platform runs without IRIS.

OptionalHIE Connector Only

↕ Enterprise migration path · Zero code changes at each stage · FHIR R4 standard maintained throughout

🚀 Zone 6 — Infrastructure Migration Path Zero code changes at every stage · Same Docker containers · FHIR R4 standard throughout

STAGE 1 · NOW

Docker on local machine

HAPI FHIR + PostgreSQL
Synthetic data only
$0/month

Running Today

Desktop: localhost:3000

STAGE 2 · PILOT

Docker on VPS ($50–80/mo)

TLS via Let's Encrypt
BAA with host provider
First health plan data

DigitalOcean / Hetzner

Config change only

STAGE 3 · SCALED

Cloud VM (AWS EC2 / Azure)

HIPAA eligible
Encrypted storage
Audit logging

2–3 health plan partners

Container migration

STAGE 4 · ENTERPRISE

Azure Health Data Services
or AWS HealthLake

HITRUST CSF
SOC 2 Type II
Full ITSM

Large contract / federal award

FHIR R4 — no rebuild

Zero-code-change guarantee: Every component is built to the FHIR R4 standard and containerized with Docker. The application code that calls http://localhost:8080/fhir in development calls the same paths at the managed FHIR endpoint in production. Only the base URL changes.

HPDM Repository Intelligence · Phase 1A Active

Statistical Foundation — Confidence Narrows As Cohorts Grow

3 Program Cohorts active · 3 Regional Validation Cohorts initializing · 6 Clinical Boundary Cohorts defined · CI width: ±0.028 rafDelta at n=3

Program Cohorts

±0.028

CI Width rafDelta

3 RVCs

Regional Validation

6 CBCs

Clinical Boundary

n=8

RVC Activation Threshold

Confidence Interval Narrowing — rafDelta 95% CI

CI width = 3.92 × (σ/√n) · σ ≈ 0.025 estimated from 3-cohort range · Current: ±0.028 at n=3 · RVC activation at n=8 (±0.017)

Statistical Strength Table

Cohorts (n)	95% CI Width	CI as % Mean	Status
3 ← Current	±0.028	24.2%	CT · IN · PA
8	±0.017	15.1%	RVC Activation
12	±0.014	12.1%	Ind. Reference
20	±0.011	9.5%	CMS-Comparable

Regional Validation Cohorts (RVCs) — Active

HRSA / USDA / HHS anchored · Activated when ≥2 contributing Program Cohorts

RVC-01 · Northeast Rural · Connecticut RHT (PC-CT-01) → contributing

RVC-02 · Mid-Atlantic Rural · Pennsylvania RHT (PC-PA-01) → contributing

RVC-04 · Great Lakes Rural · Indiana GROW (PC-IN-01) → contributing

RVC-03, 05–08 · Pending ≥2 contributing Program Cohorts

Clinical Boundary Cohorts (CBCs) — Framework Defined

Cross-state clinical/demographic logic · Crosses political geography · Activated when ≥2 contributing Program Cohorts within boundary

CBC-01 · Appalachian · ARC-defined · PA RHT contributing

CBC-02 · Mississippi Delta · DRA-defined · Pending

CBC-03 · Great Plains · USDA RUCC 7-9 anchored · Pending

CBC-04 · Northern New England · Pending

CBC-05 · Pacific Northwest Rural · Pending

CBC-06 · Gulf Coast Rural · Pending

Program Engagement Pipeline · 3 Active Programs

Engagement Status — CT · IN · PA

Active Programs

$2.511B

Total RAF Gap · All Cohorts

July 2026

Next Deadline · IN GROW RFP

Open Data Agreements

Phase 1A

Current Platform Stage

Connecticut · CT RHT

Care 4 Connecticuters · PC-CT-01

ACTIVE · PHASE 1A

$265.5M

RAF Gap/yr

APCD Pending

CT OHS OI-05

Milestones

✅ CT_cohort_v1 built · 80,826 MCC pts
✅ C4C Governance — David (MPO) · George Reigeluth (Clinical PO)
🔄 CIO Engagement — Sumit (next target)
🔄 CT OHS APCD data agreement
⬜ AHEAD Demo alignment · Connie HIE
⬜ CT_cohort_v2 (triggers Phase 1B)

BF-01: Hypersphere license + QPROI + Alpha Omega RPM + EHR costs not yet reconciled vs $25M grant

Indiana · GROW Region 8

MyTruAdvantage/SIHO · PC-IN-01

RFP JULY 2026

$310.2M

RAF Gap/yr

35/30/35

Revenue Share

Milestones

✅ IN_cohort_v1 · 100K MCC pts · 108-pg impl strategy
✅ 5 hospital partners identified
✅ Gene Cronin leads SIHO relationship
🔄 SIHO/MyTruAdvantage engagement active
🔄 GROW RFP preparation · due July 2026
⬜ IHIE (Indiana HIE) IRIS integration (OI-03)

RPM: $4.9M–$12M/yr guided (HPDM-identified via Alpha Omega). 500K patient milestone = Meeting 2 only.

Pennsylvania · RHT

Penn State Health · PC-PA-01

ACTIVE · DEVELOPMENT

$1.935B

RAF Gap/yr

2.4M

Rural Residents

Milestones

✅ PA_cohort_v1 · 1.488M MCC pts
✅ PA RHT Executive Brief v2 — Dr. Chip Mershon
✅ Steven T. Brown Jr. (QPROI bundled in license)
🔄 Penn State Health formal engagement
⬜ PA APCD data agreement
⬜ PA_cohort_v2 (triggers Phase 1B)

Scoping: Univ of Pittsburgh MRC = PA RHT independent evaluator only. Never in CT or platform-wide materials.

Delta Analytics Engine · Phase 2 · Self-Validating Loop

Score vs. Claims Validation — Proving HPDM Predicts Before Claims Show It

25.8×

Cost Separation · Imminent vs Stable

5 Tiers

Calibrated Risk Separation

$2.05M

Net Benefit · 35% Efficacy

Phase 2

Requires Real Claims

APCD

Trigger: CT OHS + IN IDOH

Cost Separation by HPDM Risk Tier — Indiana GROW (100K Calibration)

Imminent

$89,200/yr avg cost

25.8×

Critical

$64,300/yr avg cost

18.6×

High

$37,400/yr avg cost

10.8×

Moderate

$19,800/yr avg cost

5.7×

Stable

$3,460/yr avg cost

1.0×

DuckDB Score Archive vs. Claims Delta · Indiana GROW synthetic calibration · Real data integration at Phase 1B APCD trigger

ROI Model · Imminent + Critical Tiers · Indiana GROW 100K

Item	Value
Imminent + Critical (6.2% of MCC)	6,200 pts
Blended avg cost/pt/yr	$76,750
Total addressable cost	$475.9M
At 35% efficacy — reduction	$166.6M
Platform + intervention cost	−$164.6M
Net benefit (conservative 35%)	$2.05M/yr

At 50% efficacy (HPDM-guided RPM full activation): $73.4M net benefit. Delta Engine proves the efficacy assumption over time against real claims.

Delta Analytics Engine — Build Phases

Phase 1A · Active

Score Archive running. Every HPDM score timestamped → DuckDB. This creates the prediction record that Delta Engine validates. No claims data needed yet.

Phase 1B · Triggered by APCD

CT OHS APCD + Indiana IDOH execute → real claims flow into Azure SQL. Score Archive joins against Claims by patient + date. First real prediction-vs-outcome validation.

Phase 2 · Self-Validating Loop

Delta outputs feed back into HPDM weight calibration. Conditions with high prediction accuracy get higher scoring weights. Self-improving loop no competitor can replicate.