Healthcare & Life Sciences

End-to-end walkthroughs showing how a regional health system uses the MATIH Platform to unify clinical data, predict patient outcomes, optimize operations, and maintain strict HIPAA compliance across every data interaction.

Industry Context

Healthcare organizations sit on some of the richest and most sensitive data in any industry. A single patient encounter generates records across electronic health records (EHR), laboratory information systems, pharmacy dispensing, billing and claims, imaging archives, and patient-reported outcomes. The challenge is not volume alone -- it is fragmentation across dozens of siloed clinical and administrative systems, each with its own data model, access controls, and regulatory obligations.

Most health system analytics teams operate in reactive mode: pulling CSV extracts from the EHR, manually reconciling claims data, and building one-off reports for each regulatory submission. Data scientists struggle to access de-identified datasets for research. Clinicians lack real-time operational visibility. Executives receive monthly reports that are already stale. The MATIH Platform consolidates these workflows into a single governed environment where clinical researchers, operational leaders, and executive decision-makers work from the same trusted, HIPAA-compliant data layer.

Company Profile: Pinnacle Health System

All walkthroughs in this section follow employees at Pinnacle Health System, a fictional regional health system with the following profile:

Attribute	Value
Hospitals	12 acute care facilities
Annual Patient Volume	200,000 unique patients/year
Beds	3,200 licensed beds across system
Annual Revenue	$4.2B
Employed Physicians	1,800
Payer Mix	42% Medicare, 28% Commercial, 18% Medicaid, 12% Self-pay
EHR System	Epic (7 hospitals), Cerner (5 hospitals)
Data Team	6 data scientists, 3 ML engineers, 5 BI analysts, CMO, CMIO

Sample Datasets

These are the core datasets used across all four walkthroughs. In a production deployment, these tables live in their respective source systems and are ingested into the platform via Airbyte connectors, FHIR APIs, or file imports.

Dataset	Source	Rows	Description
`patients`	EHR (FHIR)	200K	Demographics -- patient_id, mrn, birth_date, gender, race, ethnicity, zip_code, insurance_type
`encounters`	EHR (FHIR)	2M	Admissions, ED visits, outpatient -- encounter_id, patient_id, admit_date, discharge_date, facility_id, discharge_disposition
`lab_results`	LIS	5M	Lab values -- result_id, encounter_id, loinc_code, test_name, result_value, result_units, reference_range, collected_at
`prescriptions`	EHR (FHIR)	1.5M	Medications -- rx_id, patient_id, ndc_code, drug_name, dosage, frequency, prescriber_id, start_date, end_date
`claims`	Claims DB	3M	Billing -- claim_id, encounter_id, payer_id, drg_code, billed_amount, allowed_amount, paid_amount, denial_code
`clinical_trials`	CTMS	500	Active trials -- trial_id, nct_number, title, phase, therapeutic_area, pi_name, status, target_enrollment
`trial_enrollments`	CTMS	50K	Enrollment records -- enrollment_id, trial_id, patient_id, consent_date, status, site_id
`imaging_metadata`	PACS	800K	Radiology metadata -- study_id, patient_id, modality, body_part, study_date, reading_physician, findings_summary

Data Sources

Pinnacle Health's data lives across clinical, administrative, and research systems. The platform connects to all of them through the Ingestion Service (Airbyte connectors), FHIR APIs, and file imports.

Source	Type	Connector	Sync Mode	Frequency
Epic EHR (7 hospitals)	FHIR R4 API	Airbyte FHIR Connector	Incremental (lastUpdated)	Every 15 min
Cerner EHR (5 hospitals)	FHIR R4 API	Airbyte FHIR Connector	Incremental (lastUpdated)	Every 15 min
Claims Clearinghouse	PostgreSQL	Airbyte PostgreSQL CDC	Incremental (WAL)	Hourly
Lab Information System	HL7v2 / FHIR	Airbyte FHIR Connector	Incremental	Every 30 min
CTMS (Clinical Trial Mgmt)	PostgreSQL	Airbyte PostgreSQL	Incremental (timestamp)	Daily
RedCap Surveys	CSV Export	File Import (Data Workbench)	One-time / on-demand	Weekly
ClinicalTrials.gov	REST API	Airbyte HTTP API	Full refresh	Weekly

Data Flow Architecture

                     Pinnacle Health System Data Flow
  ┌──────────────────────────────────────────────────────────────────────┐
  │                        CLINICAL DATA SOURCES                        │
  │  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐  │
  │  │Epic EHR  │ │Cerner EHR│ │  Claims  │ │   LIS    │ │  CTMS    │  │
  │  │(FHIR R4) │ │(FHIR R4) │ │(Postgres)│ │(HL7/FHIR)│ │(Postgres)│  │
  │  └────┬─────┘ └────┬─────┘ └────┬─────┘ └────┬─────┘ └────┬─────┘  │
  └───────┼────────────┼────────────┼────────────┼────────────┼────────┘
          │            │            │            │            │
          ▼            ▼            ▼            ▼            ▼
  ┌──────────────────────────────────────────────────────────────────────┐
  │              INGESTION SERVICE (Airbyte + FHIR Connectors)          │
  │      De-identification  |  FHIR-to-relational  |  Schema mapping    │
  └────────────────────────────────┬─────────────────────────────────────┘
                                   │
                                   ▼
  ┌──────────────────────────────────────────────────────────────────────┐
  │                 HIPAA-COMPLIANT PLATFORM DATA LAYER                 │
  │  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌────────────────────┐  │
  │  │ Catalog  │  │  Query   │  │ Quality  │  │    Governance      │  │
  │  │ Service  │  │  Engine  │  │ Service  │  │  (masking, audit,  │  │
  │  │          │  │ (Trino)  │  │ (GX)     │  │   HIPAA policies)  │  │
  │  └──────────┘  └──────────┘  └──────────┘  └────────────────────┘  │
  └────────────────────────────────┬─────────────────────────────────────┘
                                   │
          ┌────────────────────────┼────────────────────────┐
          ▼                        ▼                        ▼
  ┌──────────────┐      ┌──────────────┐       ┌──────────────────┐
  │ ML Workbench │      │ BI Workbench │       │ Agentic Workbench│
  │              │      │              │       │                  │
  │ Readmission  │      │ Operations   │       │ NL clinical      │
  │ models,      │      │ Command      │       │ queries,         │
  │ Trial        │      │ Center,      │       │ Evidence-based   │
  │ matching     │      │ Quality      │       │ decision         │
  │              │      │ dashboards   │       │ support          │
  └──────────────┘      └──────────────┘       └──────────────────┘

Compliance Framework

Healthcare data is governed by strict federal and state regulations. The MATIH Platform enforces compliance at every layer -- from ingestion to visualization.

Regulation	Scope	Platform Enforcement
HIPAA Privacy Rule	Protected Health Information (PHI) -- 18 identifiers	Column-level masking, role-based access, minimum necessary enforcement
HIPAA Security Rule	Electronic PHI (ePHI) safeguards	Encryption at rest and in transit, audit logging, access controls
HITECH Act	Breach notification, EHR meaningful use	Automated audit trails, data access reporting
FDA 21 CFR Part 11	Electronic records in clinical trials	Immutable audit trails, electronic signatures, data integrity validation
CMS Conditions of Participation	Quality reporting, readmission penalties	Automated metric computation matching CMS methodology

HIPAA PHI Identifiers -- Governance Rules

The Governance Service automatically detects and enforces masking on all 18 HIPAA identifiers. The following governance policy is applied at the platform level:

{
  "policy_name": "hipaa_phi_masking",
  "policy_type": "column_masking",
  "description": "Mask all 18 HIPAA identifiers for non-privileged roles",
  "rules": [
    {
      "identifier": "patient_name",
      "columns": ["first_name", "last_name", "full_name"],
      "mask_type": "hash",
      "allowed_roles": ["treating_physician", "hipaa_officer", "data_steward"]
    },
    {
      "identifier": "date_of_birth",
      "columns": ["birth_date", "dob"],
      "mask_type": "generalize_year",
      "allowed_roles": ["treating_physician", "clinical_researcher"]
    },
    {
      "identifier": "ssn",
      "columns": ["social_security_number", "ssn"],
      "mask_type": "redact",
      "allowed_roles": ["hipaa_officer"]
    },
    {
      "identifier": "mrn",
      "columns": ["medical_record_number", "mrn"],
      "mask_type": "tokenize",
      "allowed_roles": ["treating_physician", "clinical_researcher"]
    },
    {
      "identifier": "geographic",
      "columns": ["street_address", "zip_code"],
      "mask_type": "generalize_zip3",
      "allowed_roles": ["hipaa_officer"]
    }
  ],
  "audit": {
    "log_all_access": true,
    "retention_days": 2190,
    "alert_on_bulk_access": true,
    "bulk_threshold": 500
  }
}

Business KPIs

Pinnacle Health tracks these key performance indicators across all workbenches and dashboards. Each walkthrough shows how the platform computes, monitors, and acts on these metrics.

KPI	Definition	Current	Target	CMS Benchmark
30-Day Readmission Rate	% of discharges readmitted within 30 days	14.2%	< 12.0%	15.5% national avg
Average Length of Stay (ALOS)	Mean inpatient days per admission	4.8 days	4.2 days	4.5 days
Patient Satisfaction (HCAHPS)	Hospital Consumer Assessment scores	72/100	80/100	71/100 national avg
Clinical Trial Enrollment Rate	Eligible patients enrolled / eligible identified	8.3%	15.0%	5-10% industry avg
Claims Denial Rate	Denied claims / total claims submitted	11.4%	< 8.0%	10% industry avg
Bed Utilization Rate	Occupied bed-days / available bed-days	78%	82-88%	Industry optimal
ED Boarding Time	Time from ED disposition to inpatient bed	4.2 hours	< 2 hours	--
Mortality Index (O/E)	Observed / Expected mortality ratio	1.04	< 1.00	1.00 expected
OR Utilization	Scheduled OR minutes used / available	68%	75-85%	70-80% benchmark
CMS Star Rating	Overall hospital quality rating	3.2 stars	4.0 stars	3.0 median

Persona Walkthroughs

Each walkthrough follows one persona through all eight lifecycle stages, using real Pinnacle Health data and scenarios. Start with the role closest to yours, or read all four to see how the platform enables cross-functional collaboration.

Walkthrough	Persona	Scenario	Primary Workbenches
Data Scientist Journey	Dr. Maya Chen, Clinical Data Scientist	Predicting 30-day hospital readmissions to reduce CMS penalties	ML Workbench, Data Workbench
ML Engineer Journey	Jordan Park, ML Engineer	Building a clinical trial patient matching engine at scale	ML Workbench, Pipeline Service
BI Lead Journey	Aisha Williams, BI Lead	Creating a hospital operations command center for 12 facilities	BI Workbench, Semantic Layer
Executive Leadership Journey	Dr. Robert Kim, CMO	Using AI-assisted analysis for clinical quality strategy and CMS performance	Agentic Workbench, BI Dashboards

How the Walkthroughs Connect

These four personas work on the same data at Pinnacle Health. Their work products feed into each other:

  Dr. Maya Chen (Data Scientist)         Jordan Park (ML Engineer)
  ┌────────────────────────┐            ┌────────────────────────┐
  │ Readmission risk       │            │ Clinical trial patient │
  │ model (C-stat 0.72)    │───────────▶│ matching engine        │
  │                        │  model     │                        │
  │ Feature engineering,   │  registry  │ Ray Serve deployment,  │
  │ cohort analysis        │            │ EHR integration        │
  └──────────┬─────────────┘            └──────────┬─────────────┘
             │ risk scores                         │ match alerts
             ▼                                     ▼
  ┌────────────────────────┐            ┌────────────────────────┐
  │ Aisha Williams         │            │ Dr. Robert Kim (CMO)   │
  │ (BI Lead)              │◀───────────│                        │
  │                        │ dashboard  │ AI-driven quality      │
  │ Operations Command     │ access     │ strategy, CMS Star     │
  │ Center, Quality        │            │ Rating projections,    │
  │ dashboards             │            │ board presentations    │
  └────────────────────────┘            └────────────────────────┘

Maya's readmission risk scores power Aisha's quality dashboards and trigger care coordinator interventions. Jordan's trial matching engine feeds enrollment metrics that Dr. Kim reviews in strategic planning. The semantic layer ensures all four personas use the same CMS-aligned metric definitions.

Prerequisites

Before following these walkthroughs, ensure you have:

A running MATIH Platform instance (see Installation)
The Pinnacle Health sample dataset loaded (available in the platform's sample data catalog)
Completed the Quickstart Tutorials for the workbenches you plan to use
HIPAA-compliant environment configured (see Security)

Related Chapters

Data Ingestion -- Configuring Airbyte connectors, FHIR APIs, and file imports
Query Engine -- SQL federation across clinical and administrative sources
Data Catalog -- Metadata management, HIPAA tagging, and lineage
Pipelines -- Temporal-based clinical data orchestration
ML Service -- Model training, registry, and clinical model deployment
AI Service -- Text-to-SQL and clinical decision support agents
Security & Governance -- HIPAA compliance, encryption, and access controls

Executive Leadership Journey Data Scientist Journey