Integrations¶

6 system integration mappings that connect the Data Plane source systems to the Semantic Ontology Layer. Each integration is defined in enterprise-knowledge/integrations/ as a YAML file with field-level mappings, conflict resolution, and failure handling.

Integration Summary¶

#	Integration	Source	Target	Type	Frequency
1	EHR to Revenue Cycle Sync	EHR	Revenue Cycle Management	Bidirectional	15-min interval
2	EHR to HIS Sync	EHR	Hospital Information System	Bidirectional	Real-time ADT, 15-min other
3	Lab & Imaging Feed	LIS / PACS	Ontology Layer	Unidirectional	Real-time results
4	Clinical Notes NLP Pipeline	Clinical Documentation	Ontology Layer	Unidirectional	<60 min from dictation
5	Population & External Data Feed	Payers / Public Health	Ontology Layer	Unidirectional	Monthly batch + event-driven
6	Regulatory & Quality Feed	CMS / TJC / State	Ontology Layer	Unidirectional	Daily scan

Integration Architecture¶

flowchart LR
    EHR["Electronic Health Record"] -->|"#1 Bidirectional 15min"| RCM["Revenue Cycle Mgmt"]
    EHR -->|"#2 Bidirectional RT/15min"| HIS["Hospital Info System"]
    LIS["Lab Info System"] -->|"#3 Real-time"| ONT["Ontology Layer"]
    PACS["PACS / Radiology"] -->|"#3 Real-time"| ONT
    NOTES["Clinical Documentation"] -->|"#4 <60min"| ONT
    POP["Payers / Public Health"] -->|"#5 Monthly batch"| ONT
    REG["CMS / TJC / State"] -->|"#6 Daily"| ONT

1. EHR to Revenue Cycle Sync¶

ID: INTEGRATION_EHR_RCM_V1_0 | File: ehr-rcm-sync.yaml

EHR is source of truth for clinical data (demographics, encounters, diagnoses). RCM is source of truth for financial and billing data.

Clinical Data Sync (EHR -> RCM)¶

EHR Field	RCM Field	Direction	Transformation / Constraint
Patient.MRN	RCM_Patient.MRN	EHR -> RCM	—
Patient.Demographics	RCM_Patient.Demographics	EHR -> RCM	Name, DOB, Address, Insurance_ID
Encounter.Encounter_ID	RCM_Encounter.Encounter_ID	EHR -> RCM	—
Encounter.Admit_Date	RCM_Encounter.Service_Date	EHR -> RCM	—
Encounter.Diagnoses[]	RCM_Claim.Diagnosis_Codes[]	EHR -> RCM	ICD-10-CM; primary + secondary ordering preserved
Encounter.Procedures[]	RCM_Claim.Procedure_Codes[]	EHR -> RCM	CPT/HCPCS with modifiers
Encounter.Attending_Provider	RCM_Claim.Rendering_Provider	EHR -> RCM	NPI lookup required

Financial Data Sync (RCM -> EHR)¶

RCM Field	EHR Field	Direction	Notes
RCM_Claim.Claim_Status	Encounter.Claim_Status__c	RCM -> EHR	Submitted / Accepted / Denied / Paid
RCM_Auth.Authorization_Status	Encounter.Auth_Status__c	RCM -> EHR	Approved / Pending / Denied
RCM_Eligibility.Coverage_Status	Patient.Coverage_Status__c	RCM -> EHR	Active / Inactive / Pending verification
RCM_Eligibility.Benefits_Summary	Patient.Benefits_Summary__c	RCM -> EHR	Copay, deductible, coinsurance
RCM_Clearance.Financial_Clearance	Encounter.Financial_Clearance__c	RCM -> EHR	Cleared / Pending / Hold

Conflict Resolution: EHR wins for clinical data (demographics, diagnoses, procedures); RCM wins for financial/billing data (claim status, authorization, coverage)

Failure Handling: 3 retries at 5-min intervals; dead letter queue; alert healthcare-integration-alerts@company.com

2. EHR to HIS Sync¶

ID: INTEGRATION_EHR_HIS_V1_0 | File: ehr-his-sync.yaml

EHR is source of truth for clinical orders and patient status. HIS is source of truth for operational data (bed management, OR scheduling, staffing). ADT events are real-time; all other data syncs on 15-min interval.

Clinical Orders Sync (EHR -> HIS)¶

EHR Field	HIS Field	Direction	Transformation / Constraint
ADT_Event.Event_Type	HIS_ADT.Event_Type	EHR -> HIS	Admit / Discharge / Transfer — real-time
ADT_Event.Patient_MRN	HIS_ADT.Patient_MRN	EHR -> HIS	Real-time
ADT_Event.Location	HIS_ADT.Unit_Ward	EHR -> HIS	Real-time
ADT_Event.Attending_Provider	HIS_ADT.Attending_Provider	EHR -> HIS	Real-time
Order.Bed_Request	HIS_Bed.Bed_Request	EHR -> HIS	Isolation / Telemetry / Standard
Order.Diet_Order	HIS_Dietary.Diet_Order	EHR -> HIS	Diet type, allergies, restrictions
Order.Activity_Level	HIS_Nursing.Activity_Level	EHR -> HIS	Bedrest / Ambulatory / Fall precautions

Operational Data Sync (HIS -> EHR)¶

HIS Field	EHR Field	Direction	Notes
HIS_Bed.Bed_Assignment	Encounter.Bed_Assignment__c	HIS -> EHR	Room / Bed / Unit
HIS_Bed.Bed_Status	Encounter.Bed_Status__c	HIS -> EHR	Occupied / Available / Cleaning
HIS_OR.Schedule	Encounter.OR_Schedule__c	HIS -> EHR	OR room, time, duration, team
HIS_Staffing.Staffing_Level	Unit.Staffing_Level__c	HIS -> EHR	Nurse-to-patient ratio by unit
HIS_Supply.Consumption	Encounter.Supply_Consumption__c	HIS -> EHR	Implants, supplies linked to case

Conflict Resolution: EHR wins for clinical orders (ADT, diet, activity); HIS wins for operational/bed management data

Frequency: ADT events: real-time (<30 sec); all other data: 15-min interval

3. Lab & Imaging Feed¶

ID: INTEGRATION_LAB_IMAGING_V1_0 | File: lab-imaging-feed.yaml

Unidirectional feed from Laboratory Information System (LIS) and PACS/Radiology into the Ontology Layer. Critical results are pushed immediately; routine results follow standard SLA.

Lab Results (LIS -> Ontology via HL7 ORU)¶

LIS Field	Ontology Field	Notes
OBX.LOINC_Code	Lab_Result.LOINC_Code	Standardized test identifier
OBX.Value	Lab_Result.Value	Numeric or coded result
OBX.Units	Lab_Result.Unit	UCUM standard
OBX.Reference_Range	Lab_Result.Reference_Range	Low-High for patient context
OBX.Abnormal_Flag	Lab_Result.Critical_Flag	Critical / High / Low / Normal
OBX.Observation_DateTime	Lab_Result.Result_Timestamp	—
OBR.Ordering_Provider	Lab_Result.Ordering_Provider	NPI
OBR.Specimen_Type	Lab_Result.Specimen_Type	Blood / Urine / CSF / Tissue

Imaging Results (PACS -> Ontology via HL7/FHIR DiagnosticReport)¶

PACS Field	Ontology Field	Notes
DiagnosticReport.Modality	Imaging_Result.Modality	XR / CT / MRI / US / NM / PET
DiagnosticReport.Findings	Imaging_Result.Findings	Structured findings text
DiagnosticReport.Impression	Imaging_Result.Impression	Radiologist summary impression
DiagnosticReport.Critical_Flag	Imaging_Result.Critical_Flag	Critical / Routine
DiagnosticReport.Radiologist	Imaging_Result.Interpreting_Provider	NPI
DiagnosticReport.Study_DateTime	Imaging_Result.Study_Timestamp	—
DiagnosticReport.Accession_Number	Imaging_Result.Accession_Number	Unique study identifier

Delivery SLA¶

Result Type	SLA	Trigger
Critical lab result	Immediate push (<1 min)	Critical_Flag = Critical
Routine lab result	<15 min	Standard batch cycle
Critical imaging finding	Immediate push (<1 min)	Critical_Flag = Critical
Routine imaging report	<30 min	Standard batch cycle

Extraction Pipeline:

flowchart LR
    LIS["LIS (HL7 ORU)"] --> PARSE["HL7 Parser"]
    PACS["PACS (FHIR DiagnosticReport)"] --> FHIR["FHIR Parser"]
    PARSE --> NORM["LOINC / SNOMED Normalization"]
    FHIR --> NORM
    NORM --> CRIT{"Critical Flag?"}
    CRIT -->|"Critical"| IMM["Immediate Push to Ontology + Alert"]
    CRIT -->|"Routine"| BATCH["Batch Cycle (15min Lab / 30min Imaging)"]
    BATCH --> ONT["Ontology Layer"]
    IMM --> ONT

4. Clinical Notes NLP Pipeline¶

ID: INTEGRATION_NOTES_NLP_V1_0 | File: clinical-notes-nlp.yaml

Unidirectional NLP pipeline extracting structured clinical data from physician dictation and clinical documentation. Target SLA: <60 min from dictation to ontology population.

Source Documents¶

Source	System	Format
Physician notes	Nuance DAX / 3M M*Modal	Dictated → transcribed text
Discharge summaries	EHR Documentation Module	Structured + free-text
Procedure notes	EHR Documentation Module	Free-text with templates
History & Physical (H&P)	EHR Documentation Module	Structured + free-text
Consult notes	EHR Documentation Module	Free-text

Extracted Entities¶

Extracted Entity	Target Ontology Field	Confidence Handling
Diagnoses	Encounter.NLP_Diagnoses[]	ICD-10-CM code suggestions
Medications	Patient.NLP_Medications[]	RxNorm coded; dose, route, frequency
Allergies	Patient.NLP_Allergies[]	Substance + reaction type
Procedures	Encounter.NLP_Procedures[]	CPT/SNOMED coded
SDoH Indicators	Patient.NLP_SDoH[]	Housing, food security, transportation, social isolation
Adverse Event Signals	Encounter.NLP_Adverse_Events[]	Falls, pressure injuries, medication errors

Confidence Thresholds¶

Confidence Score	Action
> 0.85	Auto-populate ontology fields
0.70 – 0.85	Flag for clinician review; pre-populate as suggestion
< 0.70	Route to manual review queue; do not auto-populate

NLP Extraction Pipeline:

flowchart LR
    DICT["Dictation / Notes"] --> ASR["ASR Transcription"]
    ASR --> NLP["Clinical NLP Engine"]
    NLP --> NER["Entity Extraction (Dx, Meds, Allergies, SDoH)"]
    NER --> CONF{"Confidence Check"}
    CONF -->|"> 0.85"| AUTO["Auto-Populate Ontology"]
    CONF -->|"0.70 – 0.85"| FLAG["Flag for Clinician Review"]
    CONF -->|"< 0.70"| MANUAL["Manual Review Queue"]
    FLAG --> AUTO
    MANUAL --> AUTO

5. Population & External Data Feed¶

ID: INTEGRATION_POPULATION_V1_0 | File: population-external-feed.yaml

Batch and event-driven feeds from payers, public health agencies, and quality organizations for population health management and value-based care analytics.

Feed	Target Entity	Frequency	Special Rules
Payer attribution lists	Population_Panel	Monthly	Reconcile against active patient roster; flag new/removed attributions
Claims data from payer	Claims_External	Monthly (90-day lag)	Identify out-of-network utilization and care gaps
SDoH databases (ADI, SVI)	SDoH_Index	Quarterly refresh	Geocoded to patient address; Area Deprivation Index + Social Vulnerability Index
Public health data (CDC, CMS)	Public_Health_Indicator	Quarterly	Disease prevalence, immunization rates, health disparities by region
Quality measure benchmarks (HEDIS)	Quality_Benchmark	Annual	HEDIS percentile thresholds for star ratings and VBC contracts

Frequency: Monthly batch for payer data; quarterly for public health/SDoH; annual for HEDIS benchmarks; event-driven for critical public health alerts (e.g., outbreak notifications)

6. Regulatory & Quality Feed¶

ID: INTEGRATION_REGULATORY_V1_0 | File: regulatory-quality-feed.yaml

Daily scan of regulatory and quality publications with LLM-based change detection and impact assessment.

Feed	Target Entity	Frequency	Special Rules
CMS Federal Register publications	Regulatory_Update	Daily scan	CoP/billing/reimbursement changes: immediate alert
TJC/DNV standard updates	Accreditation_Standard	On publication	New/revised standards: immediate alert + gap analysis trigger
State health department regulations	State_Regulation	Daily scan	Licensure/reporting changes: immediate alert
Quality measure specification updates	Quality_Measure_Spec	On publication	Measure retirement/revision: immediate alert to quality team
CMS quality program updates (MIPS, VBP)	Quality_Program	On publication	Scoring methodology changes: immediate alert

Regulatory Change Detection Pipeline:

flowchart LR
    RAW["Regulatory Publication"] --> NLP["NLP Extraction"]
    NLP --> CLASS["Classification (CoP / Billing / Accreditation / Quality)"]
    CLASS --> IMPACT{"Impact Assessment"}
    IMPACT -->|"Reimbursement / Patient Safety / Accreditation"| IMMEDIATE["Immediate Alert + Impact Report"]
    IMPACT -->|"Administrative / Minor"| BATCH["Weekly Digest"]

Failure Handling (All Integrations)¶

All integrations share a common failure handling pattern:

Setting	Value
Retry count	3
Retry interval	5 minutes
Dead letter queue	Enabled
Alert on failure	Integration-specific alert email
Fallback	Source system value is authoritative; manual update required in target

← Back to Ontology Overview