Health Digital Twin for the clinical workflow

March Health builds a continuously updating Health Digital Twin integrated with EHR and hospital information systems—so clinicians see a synthesized, current view of the patient at the point of care.

Watch the Platform in Action

Why This Matters

Clinical workflows today face structural constraints that limit proactive care.

  • Fragmented interfaces — Clinicians operate across disconnected EHR screens; the full picture is rarely in one place.
  • Buried risk indicators — Early signals and risk factors are scattered across notes, labs, and flowsheets instead of surfaced in context.
  • Late care-gap detection — Gaps in care are often identified in retrospect rather than at the moment they become actionable.
  • Unused longitudinal data — History, trends, and prior events are not consistently synthesized into a single, current representation.
  • Reactive alerting — Many alerts fire after the fact instead of supporting proactive intervention.

The Solution: The Health Digital Twin

A continuously updated, structured representation of the patient that consolidates conditions, medications, vitals, history, and indicators—recalculated over time.

  • The twin consolidates conditions, medications, vitals, history, and monitoring indicators into one structured snapshot.
  • Monitoring signals and watchlist scoring are recalculated as new data enters the system.
  • Alerts trigger when configured thresholds are crossed, supporting earlier intervention.
  • Clinical decision support—differentials, evidence, medication and procedure recommendations with interoperable codes—is delivered in a clinician-controlled workflow.
  • Patient-anchored AI chat is retrieval-based and grounded in the twin; it does not provide autonomous treatment recommendations.

This is assistive AI—designed to support clinician judgment, not replace it. The clinician remains in control of the decision pathway.

Platform Walkthroughs

Four short demos show how the Health Digital Twin works in practice.

Digital Twin Overview — Patient Review & Background Monitoring

Snapshot view of the Digital Twin: current state, monitoring context, and background evaluation. See how a single screen surfaces the patient’s structured state for review.

Digital Twin Clinical Chat — Patient-Context Q&A & Session Export

Retrieval-based chat anchored to the Digital Twin snapshot. Patient-context Q&A and session export. No treatment recommendations are given in chat.

Digital Twin Alerts — Watchlist Indicators & AI Assessment

Watchlist indicators and AI-driven assessment. Baseline indicators, care gaps, and risk signals for clinician review.

Digital Twin Recommendations — Evidence, Safety Checks & Write-Back

Key findings, evidence-grounded differential diagnosis, and recommendations with interoperable codes: medications (RxNorm), procedures (CPT), and diagnosis mapping (ICD-10). Safety checks and write-back into the workflow.

How It Works

High-level workflow from data to decision.

  1. Data from EHR — Relevant clinical data is drawn from the EHR and hospital information systems.
  2. Structured Twin Snapshot — A structured representation of the patient is built and kept current.
  3. Clinical reasoning pipeline — The platform runs reasoning over the twin to surface findings, differentials, and recommendations.
  4. Evidence grounding (RAG) — A retrieval-augmented layer grounds outputs in indexed medical knowledge.
  5. Clinician review — Recommendations and alerts are presented for clinician review within the workflow.
  6. Decision finalization — The clinician decides; actions are documented and, where applicable, written back.
  7. Feedback into Twin — New data and decisions update the Digital Twin state for the next cycle.

Evidence & Knowledge Grounding

The platform uses a RAG-backed knowledge layer to keep reasoning traceable and clinician-readable.

  • RAG-backed layer — Retrieval augments model outputs with relevant passages from indexed references.
  • Indexed medical references — Sources include established textbooks (e.g. Harrison’s, Williams) and other curated medical content.
  • Transparent reasoning — Logic and differentials are presented in a form clinicians can follow and challenge.
  • Traceable citations — Where applicable, recommendations and findings can be traced back to source material.

We do not claim endorsement by any medical institution or publisher. The system is designed for professional use and clinical governance review.

Deployment Model

Clear separation between production and demonstration.

  • Production — The production version runs embedded within the EHR and connects to the live chart. It is designed for secure integration in hospital environments.
  • This demo — The demo is a standalone MVP. No PHI is processed. All patient data shown is synthetic. A lightweight backend supports the demonstration only.
  • Real deployment — In a live deployment, the platform connects to the organization’s EHR and operates on real data within the agreed security and governance framework.

Security & Privacy

Designed for hospital and health-system environments.

  • Privacy-first design: no PHI is used in the demo environment.
  • Built for secure integration with EHR and hospital information systems.
  • Architecture is intended for deployment in environments that require strict access control and data governance.

Specific compliance and regulatory requirements depend on your organization and deployment context.

Bring Longitudinal Intelligence Into Clinical Workflow

March Health provides infrastructure for proactive, evidence-aware care—a Health Digital Twin that stays current with the chart and supports clinicians at the point of decision.

Visit march.health