Sample GalleryHealthcareD1 — Process Intelligence Brief
Healthcare · D1 · Regional Medical CenterAnonymized — Real Engagement

D1 — Process Intelligence Brief

Five governance gaps at a large NCI-Designated Comprehensive Cancer Center with 677 licensed beds and 145,000+ patients annually. Three of the five gaps carry the highest possible priority rating. This brief is PDIO Deliverable 1 of 4.

📄 D1 Deliverable — Process Intelligence Brief · Engagement ENG-001 · PDIO Phase 1 of 4

Executive Summary

This brief addresses five governance gaps at Regional Medical Center. Three of the five gaps carry the highest possible priority rating.

The most urgent gap is chemotherapy dosing governance. Between 3% and 5% of patients carry DPYD variants that make standard fluorouracil dosing potentially lethal. Today, those genetic test results sit in the lab section of the EHR and are not flagged at the point of prescribing. A pharmacist may catch it on manual review — or may not. A single missed variant costs $500,000+ in care, litigation, and CMS quality reporting impact.

The second critical gap is treatment eligibility governance. Molecular pathology results arrive as PDF reports and are reviewed manually in tumor board. No automated cross-reference exists between a patient's tumor mutation panel (KRAS, EGFR, BRCA1/2, HER2) and the drug eligibility rules. A wrong treatment line costs $200,000+ in drug costs, lost treatment windows, and secondary adverse events.

The third critical gap is immunocompromised patient environment governance. Neutropenic patients (ANC < 500) require positive-pressure clean rooms with HEPA filtration. Today, room checks are scheduled — not triggered by real-time correlation between the patient's ANC lab values and the room's environmental conditions. A pathogen exposure during neutropenic nadir costs $50,000–$150,000 per sepsis event plus CMS HAI penalties.

Process Inventory

Each process below involves decisions currently made without a governed record of who decided what, when, on what basis, and with what confidence. The gap is not that decisions are wrong — it is that no one can prove they were right.

PROC-001

Chemotherapy Dosing Governance

Critical (5)
Type

Clinical

Frequency

Per treatment cycle

Current Owner

Attending Oncologist

Governance Gap

No automated PGx cross-reference at point of ordering. Genetic test results sit in lab section, not flagged at prescribing.

PROC-002

Immunocompromised Patient Environment

Critical (5)
Type

Safety

Frequency

Continuous during neutropenic episodes

Current Owner

Infection Preventionist

Governance Gap

No real-time correlation between ANC lab values and environmental conditions. Room checks are scheduled, not triggered by patient status.

PROC-003

Treatment Eligibility Governance

Critical (5)
Type

Clinical

Frequency

Per treatment decision

Current Owner

Tumor Board / Attending Oncologist

Governance Gap

Molecular pathology results are PDFs, not structured data. No automated cross-reference between tumor profile and drug eligibility.

PROC-004

Wildfire Smoke Response

High (4)
Type

Safety

Frequency

Event-driven (wildfire season)

Current Owner

Facility Engineering

Governance Gap

No automated correlation between NASA FIRMS data, EPA AQI, and HVAC recirculation protocols.

PROC-005

Hereditary Cancer Screening Governance

High (4)
Type

Clinical

Frequency

Per patient + annual

Current Owner

Genetic Counselor / Oncology Navigator

Governance Gap

Genetic counseling results not systematically connected to screening protocol assignments. Patients fall through cracks at care transitions.

Decision Flow — Current State

PROC-001

Chemotherapy Dosing

The attending oncologist writes the chemotherapy order. If a DPYD genotype test was ordered (not always), the result lives in the lab section of the EHR. The pharmacist may review it during order verification — but the review is manual and the PGx result is not surfaced in the prescribing workflow. If the pharmacist is busy, covering multiple units, or the test was ordered by a different provider, the result can be missed entirely. There is no system flag that says "this patient has a DPYD variant — do not administer standard-dose 5-FU." The trail breaks at the gap between the lab result and the prescribing decision.

PROC-002

Immunocompromised Environment

The infection preventionist monitors neutropenic patients and coordinates with facility engineering on clean room requirements. ANC lab values arrive in the EHR. Clean room environmental data (differential pressure, HEPA status, air exchanges) lives in the Building Management System. These two systems do not talk to each other. When a patient's ANC drops below 500, the room should shift to positive pressure with restricted access — but that shift depends on a human reading the lab, recognizing the threshold, and calling engineering. At night, on weekends, or during high census, this handoff fails.

PROC-003

Treatment Eligibility

The tumor board reviews molecular pathology reports — typically multi-page PDFs from the molecular lab. A pathologist presents the findings verbally. The attending oncologist maps the mutations to treatment eligibility (KRAS wild-type → cetuximab eligible; EGFR mutant → TKI eligible; BRCA1/2 pathogenic → PARP inhibitor eligible). This mapping is done from memory or reference materials. There is no structured cross-reference between the patient's specific variant profile and the current drug eligibility database (ClinVar, COSMIC). Mismatches are caught only if someone in the room remembers the contraindication.

Signal Environment — 10 Active Signals

Environmental Signals

Signal IDSignalSourceSamplingNormalWarningCriticalProcess
CC-004Clean Room Differential PressureBMS pressure sensorContinuous≥2.5 Pa1.0–2.5 Pa<1.0 Pa or negativePROC-002
CC-005Indoor PM2.5 (Oncology Unit)Laser scattering sensorEvery 5 min0–12 μg/m³12.1–35.435.5+PROC-004

Human Health Signals

Signal IDSignalSourceSamplingNormalWarningCriticalProcess
CC-001Absolute Neutrophil Count (ANC)EHR FHIR (DiagnosticReport)Per lab draw1,500–8,000 cells/μL500–1,000 (neutropenic)<500 (severe neutropenia)PROC-002

External Intelligence Signals

Signal IDSignalSourceSamplingNormalWarningCriticalProcess
CC-006Outdoor AQIEPA AirNow APIHourly0–50 (Good)101–150 (Sensitive Groups)201+ (Very Unhealthy)PROC-004
CC-007UV IndexEPA/NWS UV Index APIDaily forecast0–2 (Low)6–7 (High)8+ (Very High)PROC-001
CC-008Respiratory Virus ActivityCDC Wastewater SurveillanceWeeklyVery Low (<2.5)Moderate (5.2–8.0)High (>11.0)PROC-002
CC-010CMS HAI BenchmarkCMS Provider Data (NHSN)QuarterlySIR <1.0SIR 1.0–1.5SIR >1.5All

Genomics / Pharmacogenomics Signals

Signal IDSignalSourceSamplingNormalWarningCriticalProcess
CC-002DPYD GenotypeMolecular Lab + CPIC APIOnce (germline)Normal metabolizer (*1/*1)Intermediate (1/2)Poor metabolizer (2A/2A, 13)PROC-001
CC-003Tumor Mutation PanelMolecular Pathology + ClinVarOnce per specimenN/AActionable mutation detectedContraindicated therapy matchPROC-003
CC-009BRCA1/2 Germline StatusMolecular Lab + ClinVar + gnomADOnce (germline)Wild typeVUSPathogenic / Likely PathogenicPROC-005

Decision Authority Map

RoleDecision ScopeResponse WindowEscalation Path
Attending OncologistChemotherapy dosing, treatment selectionImmediate (point of care)Department Chair → Chief Medical Officer
Infection PreventionistEnvironmental governance, PPE/visitor policyWithin 1 hour of triggerHospital Epidemiologist → Chief Operating Officer
Tumor Board (Committee)Treatment eligibility for complex casesWeekly meetingChief Medical Officer
Facility EngineeringHVAC, pressure, air exchange adjustmentsWithin 30 minutes of triggerDirector of Facilities → Chief Operating Officer
Genetic CounselorScreening protocol assignment, hereditary riskPer appointment (scheduled)Oncology Navigator → Department Chair
PharmacistOrder verification, drug interaction checkPer order (queue-based)Pharmacy Director → Chief Medical Officer

Gap 1: No designated authority for PGx integration at prescribing.

The oncologist writes the order. The pharmacist verifies. The lab holds the PGx result. No one owns the step that says 'before this order is filled, confirm the patient's DPYD status.' This is a gap between three roles, not a gap in any single role.

Gap 2: No real-time environmental decision authority.

When a patient's ANC drops at 2 AM and the room needs to shift to positive pressure, who makes that call? The night-shift charge nurse? The on-call infection preventionist? The answer varies by shift, by unit, and by who is available. There is no standing protocol that says 'ANC < 500 → automatic environmental governance trigger.'

Gap 3: Tumor board authority is time-limited.

The tumor board meets weekly. Treatment decisions between meetings default to the attending oncologist, who may or may not have the molecular pathology results in structured form. Urgent cases cannot wait for the next meeting.

ERI / LPRM Configuration Recommendations

Environmental Risk Index (ERI) — Facility-Level

CC-004 — Clean Room Differential Pressure20%

Primary barrier against pathogen infiltration in neutropenic rooms. Negative pressure = immediate safety event.

CC-005 — Indoor PM2.520%

Immunocompromised patients at 10x risk. Must correlate with outdoor AQI and wildfire data.

CC-006 — Outdoor AQI10%

Leading indicator for HVAC load. Recurring refinery and wildfire exposure creates recurring spikes.

CC-010 — CMS HAI Benchmark10%

Quarterly penalty exposure metric. SIR >1.5 = bottom quartile = 1% Medicare reduction.

Reserved (Phase 2 sensors)40%

VOC, CO₂, humidity, C. diff UV-C — available after sensor deployment.

Living Patient Risk Model (LPRM) — Patient-Level

CC-001 — ANC25%

Primary trigger for environmental governance. ANC <500 = febrile neutropenia risk.

CC-008 — Respiratory Virus Activity10%

Leading indicator for PPE/visitor policy changes in immunocompromised population.

CC-007 — UV Index5%

Photosensitive chemo drugs (methotrexate, 5-FU, doxorubicin). Relevant for ambulatory patients.

CC-002 — DPYD GenotypeN/A (binary)

Not weighted — this is a pass/fail gate. Poor metabolizer = 50% dose reduction, no exceptions.

CC-003 — Tumor Mutation PanelN/A (categorical)

Not weighted — maps to treatment eligibility rules, not a continuous index.

Reserved (Phase 2)60%

Continuous vitals, wearable biomarkers, real-time EHR trending.

Threshold Recommendations

MetricWarning ThresholdCritical ThresholdBasis
ERI Warning0.45Sum of weighted environmental signals indicates moderate facility risk.
ERI Critical0.70Immediate environmental governance required. Authority notified.
LPRM Warning0.40Patient-level risk indicates heightened monitoring.
LPRM Critical0.65Immediate clinical governance required. Authority escalation.
Confidence Escalation<0.70Below this threshold, ATLAS escalates to the designated human authority rather than delivering an automated recommendation.

Cost of Inaction

RiskAnnual ExposureBasis
DPYD adverse event (1 per year, conservative)$500,000+Single Grade 4+ fluorouracil toxicity: ICU stay ($150K–$300K), litigation exposure ($200K–$500K), CMS quality reporting impact. Source: FDA Safety Communication 2020; ADE litigation benchmarks.
Wrong treatment line (2 per year, conservative)$400,000+Drug costs ($120K/year), lost treatment window (non-recoverable), secondary adverse events. Source: ASCO treatment cost data 2024.
Neutropenic sepsis from environmental failure (2 per year)$100,000–$300,000Extended stay ($50K–$150K per event), CMS HAI penalty (1% Medicare reduction for bottom quartile). Source: CMS HAI Program; HHS AHRQ data.
Wildfire smoke respiratory event (1 per year)$25,000–$75,000Respiratory admission for immunocompromised patient during AQI spike. Source: EPA ISA 2019.
Hereditary screening gap (delayed diagnoses)$100,000+Late-stage vs. early-stage treatment cost differential. Source: NCI SEER Program.

Total estimated annual exposure: $1,125,000 – $1,375,000+

This is not the cost of ATLAS. This is the cost of continuing without governed decision intelligence.

Governance Receipt Requirement

For this engagement, governance receipts will be generated for each decision triggered by the configured governance rules. Each receipt contains:

·Decision ID — Unique identifier, tamper-evident
·Timestamp — UTC, immutable after sealing
·Trigger — Which signal(s) initiated the decision
·Risk Score — ERI and/or LPRM at time of decision
·Confidence — System confidence (0.0–1.0)
·Recommendation — What ATLAS recommended
·Authority — Who owned the decision
·Human Action — What the authority decided
·Rationale — Why (in the authority's words)
·SHA-256 Hash — Cryptographic seal, chain-linked
·Patent Reference — TPP96862

Next Steps

  1. Client reviews D1 and confirms or corrects the process inventory, authority structure, and risk assessment. Specific attention to: authority gap assignments, actual incident rates (if available), and any processes not captured in the intake.
  2. Upon D1 approval → Cromtec proceeds to D2 (Governance Design Specification). D2 will specify exact governance rules, signal configurations, decision routing, escalation protocols, and receipt specifications — ready for engineering implementation.
  3. D2 delivery timeline: 10 business days from D1 approval. D2 is the technical blueprint — it defines what gets built.
← Sample GalleryD2 — Governance Design Specification →

Anonymized · Real Engagement · CROMTEC.AI · Patent TPP96862

See what ATLAS would produce for your organization.

This D1 brief represents the first of four ATLAS deliverables. Start a conversation to scope your specific use case.

Start a conversation →← Back to samples