Nvidia Corp. today will use the J.P. Morgan Health Care Conference in San Francisco to announce multiple partnerships with healthcare technology firms aimed at apply artificial intelligence to speed up workflows in tasks ranging from drug discovery to genetic engineering.

The partnerships are based on Nvidia’s concept of “AI factories,” which are data centers that transform raw data into actionable AI models, or “tokens,” which serve as intelligent tools for various applications.

The alliances are also aimed at developing more capable agents, which are digital robots that act autonomously with decision-making capabilities, often guided by goals or objectives. Agentic systems can plan, reason and execute tasks across multiple steps, adapt to changing environments and contexts, and go beyond answering questions to initiating actions.

Nvidia said AI agents can now automate clinical documentation, assist patients in finding care, and even coordinate hospital operations during emergencies, such as natural disasters.

Thousands of agents

“Nvidia has over 1,000 digital healthcare startups in our inception program alone who are developing thousands of AI agents,” said Kimberly Powell. Nvidia’s vice president of healthcare. She cited the example of clinical conversational generative AI platform Abridge Inc., which uses agents to automate clinical documentation, saving doctors up to three hours of clerical work per day.

Nvidia is joining forces with Iqvia Holdings Inc., a provider of advanced analytics, technology and clinical research services for the life sciences industry, to streamline drug and medical device development. Iqvia’s 64-petabyte proprietary anonymized data set will be used to train custom models using Nvidia’s recently announced Llama Nemotron large language models “to drive more efficient clinical trial execution and workflow transformation with AI agents,” Powell said.

An alliance with the Mayo Clinic aims to co-develop a new generation of pathology models. Mayo will train Nvidia’s DGX Blackwell systems and AI tools with over 20 million digital pathology images to create “digital twins,” computerized replicas of physical entities,  for personalized healthcare.

A partnership with biotechnology firm Illumina Inc. aims to unlock insights from genomics data by combining Illumina’s sequencing technology with Nvidia’s AI tools. Nvidia said the collaboration will democratize genomics research and expand its applications in drug discovery.

“We’re going to unlock new markets for genomics by making not just the data, but the insights more accessible and driving significant advancements in disease research and drug discovery,” Powell said.

Nvidia is also teaming up with the nonprofit medical research organization Arc Institute to create open-source biological foundation models using Nvidia’s BioNeMo natural language processing framework (pictured) for biology and clinical research. The goal is to advance the understanding of DNA, RNA and protein structures to advance biomedical research, Powell said.

“Our partnership is going to focus on developing true foundation models for biology using BioNeMo and DGX cloud, with the resulting work contributed back to the open source in BioNeMo,” she said. DGX Cloud is a cloud-based AI supercomputing service.

The conference will also see the debut of a new release of BioNeMo called GenMol, a goal-directed molecular generation NeMo Inference Microservice for virtual screening. Nvidia is also releasing Blueprints for Protein Design, which are reference workflows for developing protein-based therapeutics.

“Tens of thousands of life sciences companies, research institutes and platform companies can now integrate BioNeMo alongside traditional lab work, creating an AI drug discovery factory that the industry calls ‘dry labs,’” Powell said. “We are seeing the transformation from a wet lab discovery process towards an AI factory, dry lab and moving drug discovery.”

Image: Nvidia