Celine Dion, SPS and the Criticality of NVIDIA and Intel’s Complementary Medical AI solutions

While it’s fun to talk about competitive conflict between vendors, it is better for us and the industry when vendors work together, even accidentally, to accomplish a common goal. Two weeks ago NVIDIA showcased its RSNA initiative that is focused on vastly improving AI performance and developer support. And last week, Intel announced its move to support federated data which is critical to the effort. The two efforts not only don’t conflict, their success is largely predicated on the effectiveness of the peer effort. In other words, NVIDIA’s effort won’t be successful without federated data, and without the power of a next generation AI, federated data alone won’t get you the result the market needs. 

Let’s chat about that this week. 

Celine Dion and SPS

Like many of you, I’d never heard of Stiff Person Syndrome (SPS) before this month. It is a poorly understood sickness that can reduce life expectancy significantly and causes a severe degradation of motor skills and it appears to have forced the early retirement of Celine Dion.  

Because of the infrequency of illnesses like SPS, they aren’t well researched, and, as a result, remedies are limited and tend to focus on mitigating the symptoms of the disease with limited effectiveness and no real path to a cure. The disease is progressive, and it can make those suffering with it afraid to even leave their home let alone drive or work. In effect, it quickly reduces the quality of life before killing them. 

Curing Illnesses Like SPS

It is diseases like this where AI could have the greatest impact. AI can, autonomously, look for and identify potential remedies as we found during the initial Covid 19 crisis, impressively quickly. But as advanced as AIs are becoming, particularly Deep Learning AIs, they are dependent on massive amounts of data, much of which is proprietary, protected, and inaccessible. 

Federated data, from which all of the personal information is stripped, has proven to be an effective way to get access to the critical amounts of data that exist in the market. You need both a powerful AI focused on the problem and federated data so the AI can be properly trained in order to find the answer.  


NVIDIA’s presentation at RSNA was a comprehensive pitch for AI-powered imaging systems which would enable interpretation at scale and diagnosis of related diseases could be more accurate. But they can also show differences between patients and identify those that are either resistant to a problem or initially inexplicably getting better.  

This can not only more quickly diagnose rare illnesses like SPS and provide a path to faster mitigation that can slow or stop the illness at an early stage before much permanent damage is done, but also identify better medications to mitigate or even cure the disease depending on what is found over time.  

Partnering with a variety of medical imaging and healthcare partners, this is an aggressive effort to provide a focused, advanced AI solution that would significantly improve the accuracy and timeliness of the diagnosis and any potential remedies.  

Intel and Federated Data

Intel Labs and the Perelman School of Medicine at the University of Pennsylvania announced they had completed a massive Federated Learning effort that allowed them to make significant progress on malignant brain tumors. This initiative was pulled from 71 institutions over six continents and improved brain tumor detection by 33%.  

Getting around both the privacy laws and organizational protections for this data wasn’t trivial and wouldn’t have even been possible with federated data. The result was an impressive showcase for the approach’s significant benefits and progress. While this initial study highlighted the early detection benefits, this same data can now be used to help determine the best known and potential methods for mitigating or solving the underlying problems by helping better understand the causes and identifying instances where patients surprisingly improved, situations that could point to an eventual cure.  

Wrapping Up:

AI is becoming an increasingly important part of our lives, and Deep Learning AIs have the greatest potential. But Deep Learning AIs require massive amounts of data which is often protected by government restrictions and the proprietary nature of the companies that own it. The greatest individual benefit from this technology is in medicine where it can be applied to unusual illnesses like SPS. 

Addressing this opportunity, Intel and NVIDIA’s recent medical announcements dovetailed nicely with each other and, when the related technology and approach is used together, results should be nothing short of revolutionary. Efforts like this, regardless of where they appear, promise to improve our lives and particularly our health and safety, showcasing the extreme benefits of collaboration and cooperation between vendors for critical industries like health care.  

Competition may be fun to watch, but when vendor efforts potentially layer like this one between NVIDIA and Intel, you can get an even better significant increase in progress.