Did you miss the session at the Data Summit? See on-demand here.
Follow VentureBeat with coverage from Nvidia’s GTC 2022 event,
Digital twins play a pivotal role in the Nvidia platform for developing robots and self-driving cars. But there are many significant regulatory, technical, and privacy barriers to medical digital twin use cases. At GTC this year, Nvidia is showcasing a variety of significant advances that could lead to the adoption of digital twins in medicine.
Hardware supporting AI has matured to a stage where the AI industry needs to constantly improve, update, test and validate. “We wouldn’t be able to measure it without digital twins,” Kimberly Powell, NVidia’s VP of healthcare, told a news conference.
Major healthcare advances presented at GTC include synthetic data generation, commercial release of its Clara Medical AI platform, enhanced DNA sequencing workflow, improved pharmacovigilance capabilities, and improved drug detection tools. Advanced digital twin capabilities will ultimately take advantage of these advances to dramatically improve patient safety and support new business models in the healthcare industry.
Enhancing robotic data platforms
Nvidia is an innovative leader when it comes to new AI platforms for autonomous driving, robotic design (Isaac) and healthcare (Clara). This platform facilitates the development, deployment and continuous improvement of AI for industry domains. Each platform includes all the capabilities needed to bring data from different sources into the development environment, to train new algorithms on a scale, and then to deploy on new interfering hardware.
Healthcare data accounts for 30% of global data needs and is growing at a CAGR of 36%. Nvidia Clara streamlines AI workflow with AI training; More than 40 pre-trained models, applications and platforms across the entire data center; In standalone servers; Or integrated into medical devices. However, Nvidia has not yet announced any specific medical digital twin capabilities.
In contrast, both Drive and Isaac include digital twin capabilities that simplify workflows that test product design, AI training workflows, and performance testing of new combinations of hardware and software. For example, the drive supports full simulation environments that can artificially produce variations that reflect the effects of rain, snow, and darkness. This simulation can help train and develop the model and predict when the model will not perform well enough. Similarly, Isaac helps create and test new robotic hardware and algorithms in this simulated environment.
The digital twin capabilities in medicine are more challenging due to privacy safety, medical regulations and safety considerations. Although companies today address these concerns in a single implementation, it is difficult to measure. The combination of Nvidia’s existing toolchain and recent announcements can help address these challenges.
Healthcare announcements at GTC
- The Ultra Rapid Nanopore Analysis Pipeline (UNAP) is a new DNA sequencing platform running on a single DGX A100 that costs ગણતરી 568 to $ 183 to sequence the entire genome. He has recently helped set a world record for indexing the entire genome in four hours and ten minutes.
- Support for four startups developing AI Transformers for decision making, therapy and drug discovery using the fastest supercomputers in the UK. Transformers help teams analyze unlabelled data, which represents most medical data.
- New training framework for chemistry developed in partnership with AstraZeneca and early access to the generative model Megamolbart. A natural language processing (NLP) model that reads the text format of chemical compounds and uses AI to create new molecules. Transformer chemistry models can train chemical language models with 1 billion parameters using the Nvidia Nemo Megatron framework.
- Jensen’s new domain-specific NLP model, built on biomegatron, improves the detection of adverse pharmaceutical phenomena by 12%, for a total detection rate of 88%.
- The world’s largest clinical language model generation tool in partnership with SynGatorTron, University of Florida. It automates the creation of artificial data of healthcare data to improve AI models while protecting privacy. It can be used to create digital twins for patient records as a control group in clinical trials. Complementary GatorTron models can also enhance medical chatbots, biomedical research, clinical trial matching, and medical event detection.
- Reference design for real-time medical-grade AI computing platform scheduled for early access in Nvidia Clara Holoscan MGC, Q1 2023. It promises to industrialize AI development and comes with a 10-year software stack support agreement. Early hardware partners include ADLINK, Advantech, Dedicated Computing, Kontron, Leadtek, MBX Systems, Onyx Healthcare, Portwell, Prodrive Technologies, RYOYO Electro and Yuan High-Tech.
Powering new business models
Powell expects that current breakthroughs in synthetic data generation and improved modeling tools will also help bring digital twins to healthcare. For example, past work with Cambridge One supercomputer has helped King’s College London build a synthetic brain. This allowed them to test new algorithms and representation populations differently.
Healthcare could also adopt the same business models that innovators like Tesla have brought with the SaaS offer for autonomous driving in the automobile industry. In the traditional model, car companies would sell the car and make all their money from the direct sale of the vehicle. Tesla has come up with a new programmable AI computing platform that constantly updates and improves over time.
Similarly, healthcare companies may develop innovative enhancements to existing equipment such as MRI scanners or endoscopy robots that improve medical procedures and clinical workflows. “We need these capabilities to marry new devices,” Powell said. The extra cost to build this is small compared to the economics of the services you can build on top of it.
Powell predicts that the current task of weaving digital twins into the healthcare AI workflow is just the beginning of a journey of at least ten years. And beyond that, it will become an important part of AI and product development in healthcare.
“Everything we do today is synonymous with how pervasive AI is. Digital twins will be pervasive in everything we do in every industry over the next decade, ”Powell said.
Venturebeat’s mission Digital Town Square is set to become a place for technical decision makers to gain knowledge about the changing enterprise technology and practices. Learn more