NVIDIA, Scripps partner on AI for genomics, sensors

October 29, 2018
by Thomas Dworetzky, Contributing Reporter
Scripps Research Translational Institute and NVIDIA are teaming up to push the frontiers of AI and deep learning into both genomics and sensor technology.

They will set up a center focusing on applications in those fields, the company announced on its blog Tuesday.

AI in medicine has tremendous promise, but a lot depends on validating its algorithms – and data sensors are key to “proving clinical efficacy,” said the institute's founder, Eric Topol, adding, “the data inputs from sensors and sequencing, in particular, will play an important role.”

The current efforts will focus on a deep-learning approach to vast data sets of health information – now doubling every seven months – thanks to faster, cheaper sequencing equipment, and the growth of wearables and personal sensors like smartwatches, blood pressure cuffs and glucose monitors. The growth is so fast, in fact, that deep-learning approaches have increased their use in genomic research papers 40 times in just the last four years, according to NVIDIA.

NVIDIA has made other deals to further health-oriented research recently. Earlier in October, it partnered with King’s College London enabling researchers to deploy NVIDIA's DGX-2 IA research system and the NVIDIA Clara platform.

Kimberly Powell, vice president of healthcare at NVIDIA, said the DGX-2, with a large memory and two petaflops of computing power, allows for faster training of massive 3D volumetric data sets and enhances the use of Niftynet, which trains neural networks for medical imaging.

“It would, historically, take many weeks to do one training cycle,” Powell told HCB News. “With the DGX-2, you can cut that down to a couple of hours.”

In August, Vyasa Analytics, a deep learning software maker in the healthcare field,
For example, the software is able to integrate and analyze this complicated data to address questions in areas including business development, EHR analytics, compliance/fraud detection, crystal morphology classification for formulation, drug repurposing and de novo compound design, said the company.

“NVIDIA is a clear leader in the deep learning revolution, and we are honored to have been accepted into their renowned Inception program,” said Dr. Christopher Bouton, founder and CEO of Vyasa. “At Vyasa, we’re developing and applying deep learning analytics to high-value use cases for our life sciences and healthcare clients, and NVIDIA’s technology will help us fuel these advances.”

The evolution of NVIDIA to a player in the healthcare space has been rapid – it made its first appearance at RSNA in 2017. That appearance was the result of a decade-long effort that now has it working with major firms like GE Healthcare on new algorithms, such as one for CT dose reduction and image processing.

“We’ve never exhibited at RSNA because we’ve largely been in the instrument or in the workstation that runs the PACS software,” Powell told HCB News during an interview at the show. “Over the last couple [of years] there’s all this really interesting new opportunity that’s coming about in all the image processing techniques, brand new visualization techniques and, of course, now artificial intelligence.”

She thinks the growth into health from the company's graphic-heavy roots means there is “definitely going to be an opportunity for a new computing platform that will be needed for health care,” Powell continued. “For self-driving cars, we’ve created an AI supercomputer that goes in the car. It is about the size of a textbook and it has the same computational capacity as 60 servers. If we can understand the problem in health care from start to finish, of when the patient enters the room to be imaged all the way through to presenting the information to the radiologists, we can make all these computational efficiencies across that, and introduce all sorts of new capabilities and innovations.”