AI for Healthcare and Life Sciences

NVIDIA BioNeMo™ is the development platform for AI-driven biology and drug discovery.

The platform includes open models, libraries, datasets, and NVIDIA NIM™ microservices for the entire AI lifecycle, enabling researchers and developers to build, customize, and deploy AI applications that drive the next experiment.

Use Cases: biofoundation model building, molecular design, virtual screening, protein structure prediction, protein binder design

• 2x faster biofoundation model training
• 6x faster model inference
• Three model acceleration recipes

NVIDIA Nemotron™ is the foundation for digital health AI. 

Nemotron offers optimized open models with open weights, training data, and recipes for advanced reasoning, coding, visual understanding, agentic tasks, safety, speech, and information retrieval, delivering leading efficiency and accuracy for building specialized AI agents.

Use Cases: ambient healthcare agents, deep clinical research agents

• Open models for visual understanding, information retrieval, speech, and safety
• Five use case examples
• 16 cookbooks on how to deploy and use the models

NVIDIA Isaac™ is a robotics and simulation platform.

This healthcare robotics developer platform is centered on simulation and synthetic data, enabling design, validation, and runtime deployment. Based on NVIDIA’s training–simulation–runtime architecture, it connects AI development directly to deployment on edge robotic systems.

Use Cases: hospital automation, teleoperation, autonomous ultrasound, surgical autonomy

• Synthetic data generation pipelines for anatomy, procedural variation, hospital environments, and patient interaction
• Sensor simulation pipelines for endoscopy, ultrasound, CT imaging, and multimodal sensor fusion

MONAI is the open medical imaging AI framework.

This open source, PyTorch-based medical imaging AI framework brings together domain-optimized data loaders, rich transformers, and 50+ high-quality pretrained models to streamline everything from 2D and 3D segmentation and registration to reporting and multimodal workflows.

Use Cases: medical imaging model training, medical imaging segmentation and annotation

• Over 8 million downloads and over 4,000 projects
• Multimodal, integrating imaging, text, and agentic AI for radiology, surgery, and pathology workflows across images and clinical documentation
• 20 medical imaging challenges won

The NVIDIA Holoscan SDK is a runtime for physical AI and sensor processing.

Holoscan is a real-time runtime for building GPU-accelerated systems that ingest high-bandwidth, multimodal, multi-rate sensor data, perform AI reasoning, and drive deterministic real-world actions. With a graph-based execution engine and accelerated input and output (IO), Holoscan scales from NVIDIA Jetson Nano™ to DGX™.

Use Cases: real-time robotics and autonomous machines, AI raw-data-to-insights for medical devices, deterministic edge AI deployment

• Scale to tens or hundreds of gigabits per second (Gb/s) without extensive system optimizations
• Accelerated IO directly into GPU memory with ready-to-use, advanced network operators, NVIDIA Rivermax®, and Holoscan Sensor Bridge
• 10 minutes to inference with domain accelerator examples

NVIDIA Parabricks® enables GPU-accelerated genomics.

This scalable genomics software suite for secondary analysis offers GPU-accelerated implementations of BWA-MEM, GATK, and DeepVariant for whole-genome and exome workflows.

Use Cases: cancer and rare diseases, NICU and critical care, population studies, research

• Over 100x faster analysis of whole genome sequencing (WGS)
• 50 percent lower compute cost for WGS
• 23 accelerated tools and six pipelines