As a young girl, Karen Perry dreamed of one day becoming a surgeon so she could help people in need.
Today, her three sisters, including her identical twin, are all nurses and nurse practitioners. Perry started out as a pre-med major in college, but something shook her during an anatomy class. She noticed a gap between her own eagerness to dissect cadavers and her classmates’ queasy, emotional response to the task.
“I thought, ‘Oh my gosh, if I’m a doctor, I’ll have to deal with all these emotional people,’” Perry recalled.
So, Perry switched her major: first to biochemical engineering and then finally to computer science. Her career in technology has touched on telecommunications, aerospace and defense, and Internet of Things security.
In the past decade, she has found herself where she always thought she would be (albeit in a different role), helping healthcare organizations improve patient care. Perry spent a big portion of her career at Intel, where she was chief healthcare solution architect until July 2025.
Perry sat down with The Forecast at Nutanix’s 2025 .NEXT event in Washington, D.C., where she shared her thoughts on how modern IT infrastructure and emerging technologies like AI can improve patient care.
Slow and Steady: Modernizing Healthcare IT
Perry noted that the COVID-19 pandemic led to radical changes in healthcare IT. Previously, patient privacy concerns and red tape prevented many organizations from rolling out solutions like telehealth and video-based patient monitoring, but the extreme circumstances of the moment left them no choice.
“During COVID, when providers had to suit up in hazmat, they had to use cameras, and so now most hospitals are using video-based patient monitoring, especially in their ICUs,” Perry said. “Having a set of experts monitoring all of these patients has led to much better outcomes.”
Perry acknowledged that the healthcare sector is typically slow to adopt new technology, not only due to institutional inertia, but also for very practical reasons related to cybersecurity, patient privacy, and the presence of latency-sensitive, lifesaving systems in their environments. Still, she noted, the COVID era illustrates the ability of healthcare organizations to adopt solutions that improve the way they deliver care to patients.
“The next phase is using these analytics in clinical workflows and leveraging generative AI,” she said.
“For instance, we could use AI to create precision treatment plans based on current patient data, historical data, and lifestyle. We could connect patients to their data. There are frankly ways for doctors to make more money by offering services like chatbot-powered patient advocates that give patients more knowledge and control.”
Perry said the key is for IT providers to focus on making clinicians' lives easier rather than on the technology itself.
“Doctors don’t want to hear about technology roadmaps,” she said. “They want to hear about how they can spend less time charting and more time with their patients.”
The Right Infrastructure for the Job
Perry noted that modern IT infrastructure is critical to supporting AI applications, especially for organizations in sensitive sectors like healthcare, where these workloads are more likely to run on-premises. She pointed Trust Domain Extensions (TDX), a technology for securing virtual machines (VMs) and Intel’s Software Guard Extensions (SGX), a security capability of Intel’s Xeon servers that allows organizations to run applications in secure enclaves.
“It’s protected from any kind of virus or access to the server,” she said. “It’s a run-time security environment that you can use to train and protect AI models. That’s super interesting to hospitals, and also pharma. They are very protective of their data.”
Perry said hyperconverged infrastructure is a particular fit for healthcare organizations that need flexible, scalable IT environments but are wary of the public cloud due to latency or security concerns.
“Nutanix is a great fit with Intel,” she said. “You don’t need lots of GPUs to do federated learning. You can use Xeon with Nutanix, and you can have these secure edge servers as part of a federated AI fabric. You can get a lot further, a lot faster, because you can securely access the data needed to train AI models.”
Democratizing AI Access in Healthcare
While it is true that implementing AI in healthcare is more complex than simply querying ChatGPT, Perry emphasized that the technology is not the exclusive domain of research hospitals or healthtech giants.
“AI doesn’t have to be incredibly expensive, and it doesn’t have to be out of reach,” she said.
According to Perry, the key is finding ways to deploy AI across heterogeneous computing environments, running on various devices, including phones, laptops, and edge servers, without compromising security.
“It’s already happening on the consumer side,” she said. “AI is running on your phone, on your laptop. Now, in regulated markets, we have to track all those devices because of security. However, we should take those examples and just harden them. Make it secure, make it cost-effective, and enable it to run at the edge.”
Perry explained that edge-based AI is especially powerful in mobile clinics, ambulances and rural settings. She pointed to work in Australia, where mobile radiology labs are beginning to analyze scans on-site.
“We can do analytics right there with the patient, instead of waiting a month,” she said.
“It’s important to be able to run analytics at the edge, but it has to be secure, especially in healthcare. Software-defined infrastructure offers the security and connectivity needed to send the results of those algorithms to the folks who need to get them.”
Calvin Hennick is a contributing writer. His work appears in BizTech, Engineering Inc., The Boston Globe Magazine and elsewhere. He is also the author of Once More to the Rodeo: A Memoir. Follow him @CalvinHennick.
© 2025 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.