See the liver. Predict the future. Intervene earlier.

LiverTwin™ fuses multi-omics, ultrasound/MRI and longitudinal EHR data to generate a patient-specific digital twin for metabolic liver disease management.

Built on NVIDIA's healthcare stack: BioNeMo – NeMo – MONAI – Triton – Holoscan – Omniverse – FHIR & DICOM compatible.

The Challenge

MASLD affects over 100 million Americans and is severely underdiagnosed. Existing tools rely on single-modality data and fail to capture the complex, dynamic nature of liver disease.

Underdiagnosed

Over 100 million Americans affected, but most cases go undetected until advanced stages

Single-Modality

Current tools analyze data in isolation, missing critical cross-modal insights

No Integration

Clinicians lack integrated insights, leaving them without comprehensive decision support

What LiverTwin™ Does

Our AI platform creates comprehensive digital twins that transform liver disease management

Builds Digital Twins

Integrates multi-omics, imaging, and EHR signals to create patient-specific digital representations that capture the full complexity of liver disease

Forecasts Progression

Predicts disease trajectory and identifies intervention opportunities before symptoms appear, enabling proactive care management

Recommends Actions

Provides next-best actions with clear rationale, integrates into clinical workflows, and auto-documents notes for seamless care delivery

Why LiverTwin™ is Different

Built from the ground up for comprehensive, explainable, and scalable liver disease management

Multimodal by Design

Seamlessly integrates omics, imaging, and clinical data for comprehensive insights that single-modality tools miss

GPU-Accelerated Foundation Model

Built on NVIDIA's AI stack for real-time processing and scalability, delivering insights at the speed of care

Explainable Outputs

Provides transparent rationale and evidence citations, building clinician trust through interpretable AI

Edge + Cloud Deployment

Flexible deployment from bedside ultrasound carts to enterprise cloud, adapting to your infrastructure needs

Built on NVIDIA

LiverTwin™ runs as secure, containerized micro-services across NVIDIA's healthcare stack

Imaging

MONAI + VISTA-3D NIM – Automated liver segmentation and quantification for CT/MRI with state-of-the-art accuracy

Synthetic Data

MAISI NIM – Scanner-harmonized, privacy-preserving augmentation to improve model robustness across sites

Genomics & Single-Cell

Parabricks + RAPIDS-singlecell + BioNeMo – Minutes-level genomics pipelines and pathway-aware embeddings

Serving & Agents

NIM + Triton + NVIDIA AI Enterprise – Hardened inference with access controls, auditability, enterprise-grade support

"Why this matters: validated components, faster pilots, easier approvals."

Core Modules

Six integrated modules that work together to create comprehensive digital twins

1

Omics Encoders

Single-cell and bulk RNA/DNA encoders produce biological pathway embeddings that capture molecular signatures

2

Imaging Encoder

2D/3D ultrasound & MRI encoders extract tissue-level signatures and structural patterns from medical imaging

3

Clinical Timeline Encoder

Converts longitudinal EHR data into disease state embeddings that track patient progression over time

4

Fusion Model

Cross-attention layers unify modalities into a comprehensive patient state representation

5

Digital-Twin Simulator

Generates time-evolving liver states and counterfactuals to predict disease progression scenarios

6

Clinician Agent

Summarizes findings, drafts notes/orders, and suggests next-best actions with clear clinical rationale

How It Works

Four seamless steps from data ingestion to clinical action

Connect Data

FHIR/EHR, DICOM/PACS, secure omics pipelines

Encode

GPU pipelines convert each modality into embeddings

Fuse & Simulate

Generative model produces the digital twin and forecast

Act

Recommendations & documentation delivered back to the EHR

Architecture

Flexible deployment options for cloud and edge environments

Ingest

FHIR/EHR, DICOM/PACS, secure omics pipelines

Encode

MONAI imaging encoders; Parabricks/RAPIDS-singlecell + BioNeMo for omics; NeMo for clinical text

Fuse & Simulate

Multimodal fusion + LiverTwin™ generative simulator packaged as NVIDIA NIM micro-services

Serve

Triton-backed inference with GPU auto-scaling

Act

Clinician Agent drafts notes/orders and writes back to the EHR

Blueprints We Use

Proven, standardized workflows that accelerate deployment and reduce integration risk

Genomic Analysis Blueprint

Standardized, reproducible pipelines for cohort-scale analysis with validated performance metrics

Single-Cell Analysis Blueprint

End-to-end workflows for scRNA-seq and spatial transcriptomics with automated quality control

Using proven blueprints shortens pilot timelines and reduces integration risk for hospital IT.

Integrations & Deployment

Seamless integration with existing healthcare infrastructure

EHR Integration

FHIR R4/R5 compliance, SMART-on-FHIR app framework, Epic/Cerner/Allscripts integration

Imaging APIs

DICOM-compliant deployment with real-time data synchronization

Real-time Sync

Continuous data synchronization with existing clinical workflows

Security, Privacy & Governance

Enterprise-grade security and compliance for healthcare environments

HIPAA-Aligned Deployment

Full compliance with healthcare privacy regulations

PHI Encryption

End-to-end encryption in transit and at rest

Role-Based Access

Granular permissions and access controls

Audit Trails

Complete logging and monitoring capabilities

Federated Learning

Privacy-preserving model training options

NVIDIA AI Enterprise Support

Validated drivers, RBAC, logging and enterprise-grade support on-prem or in the cloud

Technical Validation & Benchmarks

Rigorous testing and validation across all system components

CT/MRI liver segmentation (VISTA-initialised MONAI): Dice / HD95 scores, scanner mix and site generalisation

Ultrasound triage on edge (Holoscan IGX/Jetson): per-frame latency (ms), throughput (fps) and referral-score stability

Genomics turnaround (Parabricks pipelines): wall-clock time per sample and concordance vs CPU baselines

Single-cell/spatial embeddings: batch-effect robustness and downstream fibrosis-stage classification AUC

Clinical agent reliability: citation coverage, note-draft accuracy and clinician override rate

Who Benefits

Transforming outcomes across the healthcare ecosystem

Hospitals & IDNs

  • • Earlier risk stratification
  • • Targeted screening programs
  • • Improved care coordination
  • • Reduced readmissions
  • • LiverTwin™ ships as hardened NGC containers and supports NVIDIA AI Enterprise for RBAC, patch cadence and enterprise SLAs

Payers

  • • Population health insights
  • • Cost reduction through prevention
  • • Value-based care support
  • • Risk adjustment optimization
  • • Blueprint-backed pipelines enable population-scale risk stratification with transparent methods and reproducible runs

Pharma

  • • Patient stratification
  • • Clinical trial optimization
  • • Real-world evidence
  • • Drug development insights
  • • BioNeMo-powered multimodal embeddings support digital endpoints, stratified recruitment and real-world evidence at scale

Measurable Impact

Proven results that transform patient outcomes and healthcare economics

40%

Increase in early-stage detection rates

60%

Reduction in time to hepatology referral

85%

Adherence to care pathways

$2.3M

Avoided advanced-stage costs per 1,000 patients

Ready to see a live demo?

Experience how LiverTwin™ can transform your approach to liver disease management and patient care.

Frequently Asked Questions

Common questions about LiverTwin™ implementation and capabilities

What data do you need to start?
LiverTwin™ can work with various data modalities including EHR data, medical imaging (ultrasound, CT, MRI), and genomics data. We can start with whatever data you have available and expand capabilities as more data sources are integrated.
What does deployment involve?
Deployment is flexible and can be cloud-based, on-premises, or hybrid. We provide containerized micro-services that integrate with your existing infrastructure through standard healthcare APIs (FHIR, DICOM).
Can LiverTwin™ run on-premises?
Yes, LiverTwin™ supports on-premises deployment with NVIDIA AI Enterprise for enhanced security, compliance, and enterprise support. This ensures data never leaves your environment while providing full functionality.
How do we ensure clinicians trust the AI outputs?
LiverTwin™ provides explainable outputs with transparent rationale and evidence citations. All recommendations include clear reasoning and supporting data, allowing clinicians to understand and validate the AI's decision-making process.