CareCompile · InterMedical Consulting

CareCompile Semantic

A cloud-based, semantically interoperable data platform that harmonizes information from electronic health records, claims, and public health feeds into one consistent, computable structure — exposed through HL7 FHIR APIs and ready for analytics, clinical decision support, and AI applications in rural care settings.

Aligned to Utah RHTP · LINCS 7.3 & 7.4 · Opportunity URHTP-2694-2026

Platform overview

RFGA criterion — Establish a Data Platform

This is a live, running platform — not a slide. Every figure on this page is computed in real time from the operational database behind CareCompile's ingestion pipelines. The same architecture deploys statewide: heterogeneous feeds arrive over HL7 v2 (MLLP and HTTPS), FHIR R4, and claims interfaces; a canonical semantic layer normalizes every observation to standard terminologies; harmonized data is then served back out through FHIR APIs, analytics, and decision-support applications.

371,005HL7 v2 messages ingested
66,626Discrete lab observations
83.2%Observations semantically harmonized
141Canonical concepts (LOINC-anchored)
399Source-vocabulary aliases mapped
2,280Patient records (synthetic)
178,325API transactions logged
10,586Clinical documents indexed

Figures refresh continuously (60-second cache). Snapshot generated 2026-07-28 19:06:44 EDT. All patient-level data on this platform is synthetic, produced by the MediFlow clinical simulation engine. No protected health information is stored or displayed.

The semantic model: one data dictionary

RFGA criterion — Semantic Mapping and Data Harmonization

The core of the platform is a unified data dictionary: every clinical concept has one canonical code, one display name, a primary LOINC binding, a default unit, and a reference range. Source systems never need to agree with each other — they only need to be mapped once to the canonical layer. Diagnoses bind to ICD-10-CM (81 codes loaded), procedures and billing to HCPCS (71 codes), and laboratory observations to LOINC.

Canonical codeDisplay nameCategoryPanelLOINCUnitReference range
SODIUM Sodium chemistry CMP 2951-2 mEq/L 136-145
POTASSIUM Potassium chemistry CMP 2823-3 mEq/L 3.5-5.0
CHLORIDE Chloride chemistry CMP 2075-0 mEq/L 98-106
CO2 Carbon Dioxide chemistry CMP 2028-9 mEq/L 22-29
GLUCOSE Glucose chemistry CMP 2345-7 mg/dL 70-100
CALCIUM Calcium chemistry CMP 17861-6 mg/dL 8.5-10.5
ALB Albumin chemistry CMP 1751-7 g/dL 3.5-5.0
TP Total Protein chemistry CMP 2885-2 g/dL 6.0-8.3
LDH Lactate Dehydrogenase chemistry INFLAM 14804-9 U/L 140-280
MAGNESIUM Magnesium chemistry MISC 2601-3 mg/dL 1.7-2.2
PHOSPHORUS Phosphorus chemistry MISC 2777-1 mg/dL 2.5-4.5
URIC_ACID Uric Acid chemistry MISC mg/dL 3.5-7.2
AMYLASE Amylase chemistry MISC U/L 28-100
LIPASE Lipase chemistry MISC U/L 0-60
AMMONIA Ammonia chemistry MISC
WBC White Blood Cell Count hematology CBC 6690-2 10^3/uL 4.5-11.0
RBC Red Blood Cell Count hematology CBC 789-8 10^6/uL 4.0-5.5
HGB Hemoglobin hematology CBC 718-7 g/dL 12-17.5
HCT Hematocrit hematology CBC 4544-3 % 36-46
MCV Mean Corpuscular Volume hematology CBC 787-2 fL 80-100
MCH Mean Corpuscular Hemoglobin hematology CBC 785-6 pg 27-33
MCHC Mean Corpuscular Hgb Conc hematology CBC 786-4 g/dL 32-36
RDW Red Cell Distribution Width hematology CBC 788-0 % 11.5-14.5
PLT Platelet Count hematology CBC 777-3 10^3/uL 150-400
MPV Mean Platelet Volume hematology CBC fL 7.5-11.5
NEUT_PCT Neutrophils % hematology CBC_DIFF 770-8 % 40-70
NEUT_ABS Neutrophils Absolute hematology CBC_DIFF 10^3/uL 1.8-7.7
LYMPH_PCT Lymphocytes % hematology CBC_DIFF 731-0 % 20-45
LYMPH_ABS Lymphocytes Absolute hematology CBC_DIFF 10^3/uL 1.0-4.8
MONO_PCT Monocytes % hematology CBC_DIFF 764-1 % 2-10
EOS_PCT Eosinophils % hematology CBC_DIFF 711-2 % 0-6
BASO_PCT Basophils % hematology CBC_DIFF 704-7 % 0-2
BANDS Band Neutrophils hematology CBC_DIFF % 0-5
PT Prothrombin Time coagulation COAG 5902-2 sec 11-13.5
INR INR coagulation COAG 6301-6 0.8-1.2
PTT Partial Thromboplastin Time coagulation COAG 3173-2 sec 25-35
FIBR Fibrinogen coagulation COAG 3255-7 mg/dL 200-400
DDIMER D-Dimer coagulation COAG 48067-3 mg/L FEU 0-0.5
UA_COLOR Urine Color urinalysis UA 5778-6 Yellow
UA_CLARITY Urine Clarity urinalysis UA 5767-9 Clear
UA_SG Urine Specific Gravity urinalysis UA 2756-5 1.005-1.030
UA_PH Urine pH urinalysis UA 5803-2 5.0-7.0
UA_PROTEIN Urine Protein urinalysis UA 20454-5 Negative
UA_GLUCOSE Urine Glucose urinalysis UA 25428-4 Negative
UA_KETONES Urine Ketones urinalysis UA 5797-6 Negative
UA_BLOOD Urine Blood urinalysis UA 5794-3 Negative
UA_LEU Urine Leukocyte Esterase urinalysis UA 5799-2 Negative
UA_NITRITE Urine Nitrite urinalysis UA Negative
UA_BILIRUBIN Urine Bilirubin urinalysis UA 5770-3 Negative
UA_WBC Urine WBC urinalysis UA /HPF 0-5
UA_RBC Urine RBC urinalysis UA /HPF 0-2
UA_BACTERIA Urine Bacteria urinalysis UA None
UA_CASTS Urine Casts urinalysis UA None
PH_BLOOD pH (Blood) blood_gas ABG 2745-8 7.35-7.45
PCO2 pCO2 blood_gas ABG 2019-8 mmHg 35-45
PO2 pO2 blood_gas ABG 2703-7 mmHg 80-100
HCO3 Bicarbonate blood_gas ABG 1963-8 mEq/L 22-26
BE Base Excess blood_gas ABG mEq/L -2 to +2
O2SAT Oxygen Saturation blood_gas ABG 59408-5 % 95-100
LACTATE Lactate blood_gas ABG mmol/L 0.5-2.0
FIO2 FiO2 blood_gas ABG 0.21
TROP_I Troponin I cardiac CARDIAC 10839-9 ng/mL <0.04
TROP_T Troponin T cardiac CARDIAC 65938-7 ng/mL <0.01
BNP B-type Natriuretic Peptide cardiac CARDIAC 42757-5 pg/mL <100
NTPROBNP NT-proBNP cardiac CARDIAC 33762-6 pg/mL 0-300
CK Creatine Kinase Total cardiac CARDIAC 2151-6 U/L 30-200
CKMB CK-MB cardiac CARDIAC 2157-6 ng/mL <5.0
MYO Myoglobin cardiac CARDIAC 33562-0 ng/mL 0-90
TSH Thyroid Stimulating Hormone thyroid THYROID 3016-3 mIU/L 0.4-4.0
T3 Total T3 thyroid THYROID 3053-6 ng/dL 80-200
FREE_T3 Free T3 thyroid THYROID 3051-0 pg/mL 2.3-4.2
T4 Total T4 thyroid THYROID ng/dL 0.8-1.8
FREE_T4 Free T4 thyroid THYROID 3026-2 ng/dL 0.8-1.8
HBA1C Hemoglobin A1c endocrine ENDO 4548-4 % 4.0-5.6
CORTISOL Cortisol endocrine ENDO mcg/dL 6.2-19.4
INSULIN Insulin endocrine ENDO uU/mL 2.6-24.9
IGF1 IGF-1 endocrine ENDO 1667-6 ng/mL 115-307
ACTH ACTH endocrine ENDO pg/mL 10-60
VITD Vitamin D 25-OH endocrine ENDO 14132-0 ng/mL 30-100
B12 Vitamin B12 endocrine ENDO 2132-9 pg/mL 180-914
FOLATE Folate endocrine ENDO 2284-8 ng/mL >3
ESR Erythrocyte Sedimentation Rate inflammatory INFLAM 30341-2 mm/hr 0-20
CRP C-Reactive Protein inflammatory INFLAM 1988-5 mg/L <3.0
PROCAL Procalcitonin inflammatory INFLAM 33747-0 ng/mL <0.25
FERR Ferritin inflammatory INFLAM 2276-4 ng/mL 15-150
CHOL Total Cholesterol lipids LIPID 2093-3 mg/dL <200
HDL HDL Cholesterol lipids LIPID 2085-9 mg/dL >40
LDL LDL Cholesterol lipids LIPID 2089-1 mg/dL <100
TRIG Triglycerides lipids LIPID 2571-8 mg/dL <150
VLDL VLDL Cholesterol lipids LIPID 2522-2 mg/dL 5-40
TBIL Total Bilirubin hepatic CMP 1975-2 mg/dL 0.1-1.2
DBIL Direct Bilirubin hepatic CMP mg/dL 0-0.3
ALP Alkaline Phosphatase hepatic CMP 6768-6 U/L 44-147
AST Aspartate Aminotransferase hepatic CMP 1920-8 U/L 10-40
ALT Alanine Aminotransferase hepatic CMP 1742-6 U/L 7-56
GGT Gamma-Glutamyl Transferase hepatic LFT 2324-2 U/L 9-60
BUN Blood Urea Nitrogen renal CMP 3094-0 mg/dL 7-20
CREAT Creatinine renal CMP 2160-0 mg/dL 0.7-1.3
EGFR eGFR renal CMP mL/min/1.73m2 >60
CEA CEA tumor_marker TUMOR ng/mL <5.0
CA199 CA 19-9 tumor_marker TUMOR U/mL <37
PSA PSA tumor_marker TUMOR ng/mL <4.0
AFP AFP tumor_marker TUMOR ng/mL <10
CA125 CA-125 tumor_marker TUMOR U/mL <35
BHCG Beta-HCG tumor_marker TUMOR mIU/mL <5.0
VANCO Vancomycin drug_level DRUG_LEVEL mcg/mL 10-20
DIGOXIN Digoxin drug_level DRUG_LEVEL ng/mL 0.8-2.0
PHENYT Phenytoin drug_level DRUG_LEVEL mcg/mL 10-20
VALPRO Valproic Acid drug_level DRUG_LEVEL mcg/mL 50-100
LITHIUM Lithium drug_level DRUG_LEVEL mEq/L 0.6-1.2
THEOPH Theophylline drug_level DRUG_LEVEL mcg/mL 10-20
HBVDNA HBV DNA molecular MOLECULAR IU/mL <20
HCVRNA HCV RNA molecular MOLECULAR IU/mL <15
HIVRNA HIV RNA molecular MOLECULAR copies/mL <20
CMVDNA CMV DNA molecular MOLECULAR IU/mL <137
BRCA1 BRCA1 molecular MOLECULAR Negative
KRAS KRAS molecular MOLECULAR Wild-type
TOX_AMPHET Amphetamines Screen toxicology TOX Negative
TOX_BARB Barbiturates Screen toxicology TOX Negative
TOX_BENZO Benzodiazepines Screen toxicology TOX Negative
TOX_CANNAB Cannabinoids Screen toxicology TOX Negative
TOX_COCAINE Cocaine Screen toxicology TOX Negative
TOX_OPIATES Opiates Screen toxicology TOX Negative
TOX_PCP Phencyclidine Screen toxicology TOX Negative
TOX_METHAM Methamphetamines Screen toxicology TOX Negative
TOX_ETHANOL Ethanol toxicology TOX mg/dL <10
EKG_RATE Heart Rate (EKG) ekg EKG 8867-4 bpm 60-100
EKG_RHYTHM Heart Rhythm ekg EKG 8626-7 NSR
EKG_AXIS Heart Axis ekg EKG 8625-9 degrees -30 to +90
EKG_PR PR Interval ekg EKG 8637-4 ms 120-200
EKG_QRS QRS Duration ekg EKG 8633-3 ms 80-120
EKG_QTC QTc Interval ekg EKG 8639-0 ms 350-450
EKG_ST ST Segment ekg EKG 8630-9
EKG_INTERP EKG Interpretation ekg EKG 8634-1
EKG_COMPARE EKG Comparison ekg EKG 8631-7
KI67 Ki-67 Index pathology IHC 29593-1 % <15
HER2 HER2 pathology IHC 48676-1 score 0-1+
ERSTATUS Estrogen Receptor pathology IHC 40556-3 % N/A
PRSTATUS Progesterone Receptor pathology IHC 40557-1 % N/A
PDL1 PD-L1 TPS pathology IHC 85147-3 % <1
MSI MSI Status pathology IHC 81695-9 MSS

141 canonical concepts across 18 clinical categories (hematology 18, chemistry 15, urinalysis 15, ekg 9, toxicology 9, blood_gas 8, and more). The dictionary is an open catalog: machine-readable via the platform API and versioned under stewardship review.

Harmonization across source systems

RFGA criterion — Semantic Data Harmonization Across EHR Systems

Different EHR vendors name the same test differently — GLU, GLUC, Glucose Lvl — and semantic drift like this is exactly what breaks cross-facility analytics. The platform maintains an alias table (399 mappings, 102,718 runtime hits) that translates each source vocabulary into the canonical layer at ingestion time, so downstream consumers only ever see one vocabulary.

Coverage by source system

SourceObservationsHarmonizedCoverage
HL7 62,461 55,421 88.7%
BACKFILL-CBC-IDX 4,072 0 0%
(unlabeled) 75 0 0%
MediFlow 18 0 0%

Most-exercised alias mappings

Source aliasCanonicalLOINCHits
INR — Inr INR 6301-6 1,541
LDH — Ldh LDH 14804-9 1,376
CO2 — CO2 CO2 2028-9 1,303
BUN — BUN BUN 3094-0 1,303
PTT — Ptt PTT 3173-2 1,302
BNP — BNP BNP 42757-5 1,302
AST — Ast AST 1920-8 1,298
ALT — Alt ALT 1742-6 1,298
PT — Pt PT 5902-2 1,284
COLOR — Color UA_COLOR 5778-6 1,280
CLARITY — Clarity UA_CLARITY 5767-9 1,280
GGT — Ggt GGT 2324-2 1,279
TROP — TROP TROP_I 10839-9 1,274
CKMB — CKMB CKMB 2157-6 1,274
MYOG — MYOG MYO 33562-0 1,274
VLDL — VLDL VLDL 2522-2 1,274
WBC — WBC WBC 6690-2 1,266
CHOL — CHOL CHOL 2093-3 1,266
HDL — HDL HDL 2085-9 1,266
LDL — LDL LDL 2089-1 1,266

Overall harmonization: 55,421 of 66,626 observations (83.2%) resolve to canonical concepts. Unharmonized residue is quarantined into the orphan-concept queue below — nothing is silently dropped.

Orphan concepts: measured, queued, resolved

RFGA criterion — Semantic Data Model Evaluation

A semantic model is only trustworthy if it is honest about what it failed to map. Every observation that cannot be resolved to a canonical concept is written to a dedicated orphan-concept queue with its source code, name, sample value, and unit. The platform currently tracks 7,061 unresolved observations across 73 distinct orphan concepts, and the AI mapping assistant has generated 110 candidate LOINC bindings that await human steward approval — proposals are never auto-applied.

Source codeSource nameUnitOccurrences
5821-4 Urine WBC /hpf 1,238
13945-1 Urine RBC /hpf 1,238
11524-6 EKG 12 Channel Panel Waveform 1,030
600-7 Sputum Culture Result 744
2888-6 Urine Protein mg/dL 733
8601-3 EKG Impression 321
18843-5 Cardiac Rhythm 314
8066-1 Anticardiolipin IgG GPL 121
30504-8 Anticardiolipin IgM MPL 121
53767-6 Anti-B2 Glycoprotein I IgG U/mL 121
53768-4 Anti-B2 Glycoprotein I IgM U/mL 121
18748-4 Diagnostic imaging study 106
121022 Study Instance UID 104
27810-4 Antithrombin III Activity % 87
27811-2 Protein C Activity % 85
27816-1 Protein S Free Antigen % 81
18282-7 Cannabinoids (THC) 74
20408-1 Cocaine metabolites 73
18390-8 Opioids 73
85611-4 Lupus Anticoagulant (dRVVT Screen) 53
35492-8 MRSA Screen (nasal swab) 37
19005-8 Ultrasound Abdomen Impression 27
18907-2 Sensitivity - Vancomycin µg/mL 26
21668-8 Factor V Leiden Mutation (R506Q) 23
51309-8 JAK2 V617F Mutation 16

This queue is the evaluation report the RFGA calls for, generated continuously rather than annually: percentage mapped, the exact concepts that failed, their frequency, and a human-governed resolution workflow that feeds new dimensions back into the dictionary.

Interoperability: standards in, standards out

RFGA criteria — Interoperability · Data Infrastructure

The platform speaks the standards rural facilities already have. Inbound: HL7 v2.x over MLLP (port 2575) and HTTPS bridge, and FHIR R4 REST. Outbound: harmonized data is exposed through FHIR APIs suitable for state HIE connectivity, clinical partners, and patient access applications. Every transaction is logged with latency and outcome for the automated audit trail.

HL7 v2 feed composition

Message typeEventMessages
ADT A03 302,615
ORU R01 27,823
MDM T02 18,336
DFT P03 3,826
ORM O01 2,500
ADT A04 2,334
ADT A60 2,184
ADT A01 2,126
OMI O23 1,688
BAR P01 1,153
VXU V04 1,146
PPR PC1 1,146

FHIR R4 transaction log

ResourceTransactionsAvg latencySuccess
Observation 6 22 ms 83%
Patient 4 34 ms 100%
Encounter 2 25 ms 100%
Bundle 1 ms 100%
Condition 1 23 ms 100%
DiagnosticReport 1 47 ms 100%
MedicationAdministration 1 21 ms 100%

Live endpoints: carecompile.com:8081/api/fhir/ (FHIR R4), carecompile.com/hl7/http_bridge.php (HL7 over HTTPS), MLLP listener on port 2575.

Because applications sit on the semantic layer rather than on any single vendor's feed, new clinical, operational, and public health tools deploy rapidly: the same canonical data already powers live modules for physician decision support, infection surveillance, antibiotic stewardship, discharge forecasting, prior authorization, and care-team messaging on this infrastructure.

Value-based care and quality reporting

RFGA criterion — Value-Based Care and Quality Reporting

HEDIS-style measures stop being a manual chart-abstraction project when every result is already computable. The measures below are calculated live, right now, by joining the harmonized laboratory layer against ICD-10-coded diagnoses — the same query pattern a rural hospital would use for payer quality reporting, with zero manual effort.

MeasureDefinitionNumeratorDenominatorRate
Comprehensive Diabetes Care: HbA1c testing Patients with an active diabetes diagnosis (ICD-10 E10/E11) who have at least one harmonized HbA1c result on record 44 155 28.4%
Comprehensive Diabetes Care: nephropathy monitoring Diabetic patients with at least one harmonized renal function result (eGFR or creatinine) 64 155 41.3%
Cardiovascular screening: LDL cholesterol Patients with at least one harmonized LDL result out of all patients on the platform 41 2,280 1.8%

Measures computed against synthetic population data at page load. In production, measure specifications are versioned in the open catalog and executable against any facility connected to the semantic layer.

Continuous governance and security

RFGA criteria — Continuous Governance and Security · Cloud Architecture

Stewardship, not just software

Model changes follow a governed lifecycle: the AI mapping assistant proposes, a human steward approves, the dictionary versions, and the changelog records it. 110 terminology proposals are currently in the review pipeline; none can reach the canonical layer without sign-off. The platform operates under a published clinical workflow governance framework that defines decision tiers, mandatory human signature on clinical actions, and nine hard-stop safety gates.

Standards are documented in open catalogs: the data dictionary, alias maps, and measure definitions on this page are all machine-readable through the platform API — the transparency report is the platform itself.

Automated audit posture

API transactions under audit logging178,325
FHIR transactions logged (latency + outcome)16
FHIR transactions flagged for review (4xx/5xx or error)1
Administrative audit-log entries871

Transport security: TLS 1.2+ on all public endpoints (Let's Encrypt, auto-renewed), ModSecurity WAF with OWASP CRS, UFW host firewall, key-only SSH, role-based access control in the application layer, and systemd-supervised services with uptime monitoring.

Onboarding a rural facility

RFGA criterion — Ecosystem Onboarding and Scalability

Connecting a critical-access hospital should be a checklist, not a project. The standardized onboarding path used by this platform:

  1. Connectivity. Facility interface engine points at the platform's MLLP listener or HTTPS bridge; FHIR-capable systems register an OAuth client against the R4 endpoint. No VPN hardware required — TLS end to end.
  2. Message profiling. The first 48 hours of traffic run in profiling mode: every observation code the facility sends is inventoried automatically and compared against the canonical dictionary.
  3. Alias mapping. Codes that match known aliases harmonize immediately. Unknowns land in the orphan-concept queue, where the AI assistant proposes LOINC bindings and a terminology steward approves them — typically a single working session per facility.
  4. Validation. Side-by-side reconciliation reports confirm counts, units, and reference ranges between the source system and the semantic layer before go-live.
  5. Go-live and support. The facility appears in coverage dashboards (as on this page), with centralized technical assistance and change-management support from the stewardship team.

The same pattern already runs in production here across HL7 v2 device feeds, FHIR posters, and batch backfills — each visible as a distinct source system in the coverage table above.

Reference architecture

RFGA criterion — Cloud Architecture and Infrastructure Design

Open-source end to end: Linux, Apache, MySQL/MariaDB, PHP services, systemd-supervised listeners and queue workers, with vector search (Qdrant) and locally-hosted or cloud LLMs for the AI tier. HIPAA-aligned controls at every layer; no proprietary lock-in anywhere in the stack.

SOURCES Rural hospital EHRsHL7 v2 · MLLP / HTTPS Clinics & labsFHIR R4 REST Claims & billingDFT · X12 · HCPCS Public health feedsELR · surveillance INGESTION MLLP listener FHIR gateway Parse + validate Async queue SEMANTIC LAYER Canonical dictionary Alias mapper Orphan-concept queue Steward review + AI assist HARMONIZED STORE Computableclinical record LOINC · ICD-10 HCPCS · RxNorm Encrypted at rest Replicated off-site CONSUMERS FHIR APIs State HIE Analytics / VBC CDS + AI apps Patient access CROSS-CUTTING Security & complianceTLS 1.2+ · WAF (OWASP CRS) · RBAC · key-only SSH · audit logging on every transaction Operationssystemd supervision · uptime monitoring · nightly backups · off-site DB replica Governanceversioned dictionary · open catalogs · changelog · human sign-off on model changes

Machine-readable statistics for everything on this page: /api.php?action=overview · dictionary: /api.php?action=dictionary · orphan queue: /api.php?action=orphans