Medical Data Infrastructure

Auditable Backend for
Adaptive Clinical Trials

Modern trials are moving from centralized hospitals into patients' homes. Unify dynamic patient schemas, wearable time-series metrics, medical imagery, and immutable audit logs in one HIPAA-ready platform.

Clone Clinical Trial Blueprint Read the Docs

The Tension: Protocol Agility vs. Regulatory Rigidity

"Adaptive Trials" change their protocol mid-stream based on early findings, but healthcare software is notoriously hard to modify. You need a database that can evolve its schema in minutes, without ever breaking the strict audit trails demanded by regulators.

✗ Standard healthcare databases require weeks of engineering to add a new biomarker field.
✗ Wearable heart-rate data (metrics) is stored separately from patient demographics (dimensions).
✗ Isolating PII from data scientists usually requires complex data duplication pipelines.
✗ No automatic, tamper-proof logs to track when a doctor alters a dosage.

The Decentralized Trial Blueprint

Omnismith's headless engine supports 9 distinct data types. Build your exact protocol requirements dynamically—from numerical dosages, to blood-test PDFs, to MRI scans, to continuous biometric streams.

✓ Dynamic Templates: Add new biomarkers or lab result fields without database migrations.
✓ Rich Data Types: Natively store medical Images and lab report Files directly on the patient record.
✓ Continuous Metrics: Ingest high-speed biometric data from Apple Watch or Garmin APIs.
✓ Relational Entities: Link specific remote sensors to individual patients.

Template: Trial Participant

Patient Name String (PII)

Contact Phone String (PII)

Current Dosage (mg) Number

Blood Test Results File Upload

Baseline Scan Image Upload

Template: Remote Wearable Device

Device ID String (Unique)

Assigned Patient Reference → Trial Participant

Heart Rate (BPM) Metric (Time-Series)

Blood Glucose Metric (Time-Series)

Attribute-Level PII Isolation

Keeping data scientists away from Personally Identifiable Information is critical. Omnismith goes beyond module-level permissions by offering Resource-Scoped Permissions at the exact attribute level.

✓ Module Scopes: Allow users to view and edit entities, but disallow module settings.
✓ Template Scopes: Grant access to 'Wearables' but deny access to 'Participants'.
✓ Attribute Deny Lists: Give analysts view-access to a patient's dosage and lab files, but strictly deny access to the 'Name' and 'Phone' attributes.

Analyst Role Configuration

Participant Name

Template: Trial Participant

DENY

Contact Phone

Template: Trial Participant

DENY

Current Dosage (mg)

Template: Trial Participant

VIEW

Immutable Medical Auditing

Omnismith automatically generates a permanent ledger for every single entity change. Prove exactly when a physician authorized a new dosage, or when a lab technician uploaded the latest blood results.

✓ Meets the rigorous audit trail requirements of modern medical software.
✓ Track attribute changes, file uploads, and relationship mapping.
✓ Filter historical changes by physician, date, or specific biomarker.

Patient Audit Ledger

dr.smith@hospital.org 2 hours ago

Changed Current Dosage (mg): 15 → 25

lab-integration-api 1 day ago

Changed Blood Test Results: empty → blood-panel-oct.pdf

dr.smith@hospital.org 2 days ago

Changed Baseline Scan: empty → mri-scan-001.png

Ingest Telemetry Directly from Wearables

Stream high-frequency metric data from smartwatches, continuous glucose monitors, or bespoke remote sensors directly into Omnismith using the optimized metrics endpoint.

# JS: Webhook receiving Apple Watch HealthKit data

async function ingestBiometrics(wearableId, bpm, glucose) {
  const url = `https://api.omnismith.io/v1/entities/${wearableId}/metrics`;
  
  await fetch(url, {
    method: 'POST',
    headers: {
      'Authorization': 'Bearer omni_integration_token_xxx',
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      attribute_values: [
        { attribute_id: "heart_rate_bpm", value: bpm },
        { attribute_id: "blood_glucose", value: glucose }
      ]
    })
  });
  // Returns 202 Accepted
}