MobiHealthNews sat down with Nvidia’s VP of Health Kimberly Powell to talk AI, radiology and what health tech can learn from self-driving cars.
As more big names in tech are jumping into the healthcare arena, the industry is getting a makeover. Now Nvidia, a hardware company best known for creating consumer and professional-grade graphics processing units (GPUs) as well as chip units for mobile computing, sees an opportunity to carve out a bigger name for itself in the space.
“I think that computational biology is the future,” Kimberly Powell, VP of health at Nvidia, told MobiHealthNews during a one-on-one at the GPU Technology Conference yesterday. “I believe our legacy could really be making that contribution to creating the computing platforms for everyone. This is going to take an army. It’s not one company that is going to solve the problem, it’s not one institution who can solve the problem, and that is Nvidia’s contribution — enabling a whole ecosystem to transform things.”
Nvidia is no stranger to the healthcare industry. Earlier this week the company’s CEO Jensen Huang unveiled Clara AI, a toolkit for radiologists that includes 13 classification and segmentation artificial intelligences and software tools. This Clara AI is announcement is an expansion of the broader Clara platform, which uses AI to create a virtual medical imaging platform and was announced at last year’s conference.
“Clara AI is the next release in the Clara platform. Last year at GTC we announced project Clara. At the time we knew there was this initiative that was starting with instruments becoming more and more software defined by nature because they need to do a lot of computing and now they want to introduce AI workloads,” Powell said. “What Clara AI allows us to do is go end to end from building to managing and deploying [AI]. It is really purpos built for radiologists.”
While the company has been involved with healthcare for 11 years, within the last few it has greatly increased its efforts, particularly in AI and radiology.
“AI hit the world in 2012 and even then some of the early adopters were in medical imaging and pathology,” Powell said. “So it has taken these last several years to build up an AI platform that is accessible and democratized to the world. And we know it is incredibly useful for imaging.”
The company started looking for the best ways to implement and use AI in the healthcare space. One of the main focuses was brining radiologists into the fold and trying to meet the industry-specific needs.
“We get to use the power of all of Nvidia because we created this AI computing platform … but now we need to bolt on some of this domain-specific [information] to make it useable to non-experts in the field. This is really what this is about, [it’s] democratizing it,” she said. “Getting the radiologist involved, allowing every practice of radiology — which differs from [intistution to institution] — to build their own models to address their own problems or opportunities, to improve operations or patient care in their institutions.”
Moving into the healthcare industry has its own set of challenges — including understanding the regulatory process and implementing new technologies in such an established field, she said.
“These clinical partnerships have been absolutely critical. It is really critical to have three major parties at the table when you are thinking about this stuff,” she said. “One is to have a technology company like Nvidia, another is the clinical side of things asking ‘What does it look like in the hospital? Who is this technology or feature for?’ Finally you [need the] healthcare companies that are going to deliver some of those solutions in some way.”
While the company started in radiology, that’s not the only place it has its sights set for the future.
“There are some great things brewing in genomics,” Powell said. “Deep learning in genomics research is on the same sort of acceleration curve for deep learning overall, which means what happens in research is going to start translating into the rest of the world. I think AI and genomics is going to be a concentration moving forward, and a real opportunity.”
Drug research and development is another “absolutely exploding” avenue where Powell said the company’s technology could contribute, with the goal of improving clinical trials and shortening the time it takes for drug development.
Healthcare isn’t the only industry that the hardware company is looking to disrupt — it has also become heavily involved in self driving cars. While healthcare and cars might not seem like a natural parallel, Powell said the two can learn from each other.
“I think we can see a lot of future needs in healthcare that we have already addressed in transportation,” she said. “Healthcare is sort of where we were five years ago with self-driving cars. We are getting algorithms to market, but they are kind of static and aren’t learning all the time. We know in self-driving cars we have to be learning all the time. So they are constantly collecting data.
As such, looking at the selfdriving car industry could be an indication of the company’s next move in healthcare.
“We need those same systems need to be put in place in healthcare in the future, and that requires computing infrastructure. So we can see that need in the future and we are already thinking through those problems of post-market validation and continuous learning,” Powell said. “I could imagine us thinking through that in the next 12 to 24 months, and that [we] will be heavily involving other bodies of the ecosystem like the FDA or other people that recommend how to deploy technology into these areas of healthcare.”
Date: March 25, 2019