Artificial intelligence is all the rage in healthcare as companies look for tech-driven ways to cut costs and promote patient health. Tech giants like Intel, Google, Amazon, Microsoft and Apple have swooped in to assist payers and providers with their efforts to join the fast-paced environment.
Santa Clara, California-based Intel boasts partnerships across myriad sectors in healthcare. For example, earlier this year, not-for-profit integrated health system Sharp HealthCare, which is based in San Diego, used Intel's predictive analytics capabilities to alert its rapid-response team to identify high-risk patients before a health crisis occurred. And currently, Intel is working with pharmaceutical company Novartis on deep neural networks to accelerate content screening in drug discovery.
Healthcare Dive sat down with the head of Intel's health and life sciences business, David Ryan, to learn more about Intel's efforts to push tech- and AI-adoption across the industry.
This interview has been lightly edited for clarity and brevity.
HEALTHCARE DIVE: Intel doesn't make products per se. Tell me a bit about how Intel's life sciences and healthcare division works.
DAVID RYAN: Intel works to make the technology that's inside of all the healthcare company's products in life sciences, laboratory sciences, healthcare delivery in the acute setting, in the community, in the home. The space of my business is the space of all devices and supporting services that are related to being healthy, staying healthy, discovering and curing disease.
How does this differ across healthcare versus other industries?
RYAN: This is happening across sectors, so it's not just a healthcare topic. AI is a information age topic. What we've seen happen in healthcare is that until just several years ago, the health delivery systems did not have giant reservoirs of digital records and information. But what's been being worked on over the past 15 years is to get the information off paper and into a digitized format. So while we're not completely done with digital or digitized healthcare — it's not 100% by any stretch of the imagination — but the lion's share of the system and the lion's share of providers are now digital.
So what that has meant is that, until maybe five years ago, the giant pools of data that a hospital system would have that they were storing would be medical images, X-rays and CT scans and MRIs. Now, all forms of data related to the patient workflow, the administration of the system as well as the diagnosis and the lab tests and the interaction between all of the providers — the entire flow is the instrument and it's all being captured in data. And so the quantity of data has become just so large that no individual analyst could really analyze it, nor can you really get your arms around what's the insight inside that data. You can't write the algorithms or write the rules and achieve anything that's really very effective. So AI in healthcare, while it's been in the research divisions of the healthcare system organizations for many years, over the last several years we've seen it jump the fence. At companies that make systems, devices that make solutions that either run in the cloud or in devices in the infrastructure or at the edge like X-ray and cat scan and MRI machines, AI is now flowing out.
How would you categorize the state of AI right now?
RYAN: It is now a topic in just about every technology, every conversation we have with makers of other life science products and solutions. Now, over the last year, the first FDA-approved AI functions have released. There's probably about 50 of them. The most recent one happened just last week. GE Healthcare has an AI embedded X-ray that was approved, I think, last week. And what the algorithm does — and we were lucky enough to work with them for some time on this — what it does is identifies, in a lung X-ray, a pneumothorax in the lung. That's a potentially collapsed lung.
And that's important because, inside the machine as the X-ray is produced, you get the analysis right then in near real time as to whether there's a concern about collapsed lung with that patient. Now, good technicians would notice that. But you can't say that every technician notices it. So patients might just be X-rayed before they're sent home from the hospital and two, three, four days, a week later when the X-ray's looked at, only then is the collapsed lung noticed. It's likely that, by that time, the patient's died or they've bounced back to the hospital in an ambulance.
But by doing that complicated analysis, AI gets the information to the technician that says, 'Hey, this image should be looked at immediately by a radiologist before the patient is discharged.' And then they can intervene and deal with the condition and it could save people's lives. This journey over the last three to five years has been to work with companies on taking AI out of the research realm and into the product and real world realm of the infrastructure of healthcare and the edge of care. And it's pretty cool to see over the last year now the FDA approvals of these functions starting to rack up.
Radiology seems the most obvious area for AI adoption in a clinical setting. How should providers and tech companies walk the fine line between AI as an assistant to the clinician versus acting as the clinician itself?
RYAN: You're right about radiology. Because of the size of the data, because of the size of images, that was the first giant data that hospitals were dealing with. And so yes, AI being used on images and to help unburden the radiologist from looking at all these normal images and helping them zero in on the abnormal images more quickly and triaging the images. That's a first application area.
Regarding AI being used as a more autonomous diagnostic tool, if we leap to emerging markets, health in countries where there are no doctors or nurses for a hundred miles — that's a different conversation. The fundamental paradigm of what the industry is doing is using AI to get better algorithms for supporting the clinician. And in a few different categories, too: not only interpreting the medical information, the biomedical information or the physiological information, but also workflow and triaging things and work order and work patterns, the positioning of a patient prior to a scan — it takes a fair bit of time, and there's solutions being worked on that speed that up using AI.
For treatment, if you zoom out from the individual patient to a population of people that are in a hospital in any given day, prediction and early intervention to prevent respiratory distress, acute kidney injury, sepsis. Then there's the people who are at home: telemedicine, remote monitoring. The rate of deployment of tele-solutions is rocketing. It's in the 40% to 50% annual growth rate, though it started from a very low base. You can scale the remote solutions pretty effectively through maybe the first few thousand patients. But if you want to think about everybody, on a giant scale or an entire population, having access to remote monitoring in particular, you have to have much stronger analytic support of the folks who are managing those cases. So AI is going to be needed to scale remote care.
Are there any particular partnerships you're most excited about currently or are excited about building out?
RYAN: All of those examples. Everything we do is helping customers achieve what they're trying to do. Related to remote care, there's a category of companies that make devices for those types of solutions. Last time I looked, we were working with about 120 of them. But generally, we help companies connect with other parts of the industry that they need to be working with to make a complete solution. We'll help them architect their solution and we'll be a supplier of the chips and the technology that they build their solution with.
An example is InTouch Health, a company that makes video conferencing screens for telemedicine. They're all over the place, used by many different providers for many different use cases. We're proud to have InTouch as a customer that's really driving innovation in telemedicine.
Do you find that traditional healthcare companies are wary of partnering with the 'Big Tech Company'?
RYAN: Depends on what healthcare company it is — and depends on what tech company you're talking about. We do work with the end customers: hospitals, research institutes, the ministries of health, pharmaceutical companies, insurance companies. All of the end where healthcare lands on the ground, the paying for it, the delivering it, doing the research for it — we work with and have strategic relationships with those organizations.
We're not selling anything to them. We're not trying to take down a purchase order by meeting with the CIO of a hospital. What we are doing when we're meeting with the C-suite of these end customers is answering their questions about, where's technology going? What's happening in the industry? Does anybody have a great example of doing X? Because we have a global footprint and we are working with and supporting the majority of the ecosystem that's providing solutions.
We're able to create connections for a hospital or a pharmaceutical company of the availability of solutions from some innovative companies that they may not know about. So if you think about a device or a computer itself, every company around it and downstream from that, the nature of their relationship with Intel is not a commercial one. It's a strategic collaboration to try and help a new solution be realized more quickly.
Intel's best known for its computer chips. Is its foray into healthcare something that you see Intel being more bullish on in the future?
RYAN: We've been working in the health, life science space for about 20 years — it's not new. Our very first efforts were ethnographic and social social science research efforts to try and understand how the industry behaves. What clinicians do, how they work, what they care about, the problems they face. We began 20 years ago studying the human factors of healthcare and trying to get a point of view. And we've been continuing along that path ever since. It's not the only sector of the global economy that Intel works with strategically like this — it's one of the top five, along with financial services and other large digitally-transforming sectors. For a long time, we've have teams and businesses focused on working with these sectors strategically to architect, achieve and digest digitalization and digital transformation.
Are there other areas of research or areas of partnership that Intel hasn't tapped into yet that you'd be interested in pursuing over the next five years to a decade?
RYAN: I've been in this space 11 years. I would say about five years ago, if you'd asked me that question, I might be a little frustrated because of the speed of the end market. We're a Silicon Valley computer company. We move at chip speed, which is 10 times faster than Internet speed. But that's really changed. The energy and the passion in our folks who work in this space is really strong. We're seeing the market we're working with here and the companies we're working with — they're moving so much more quickly than just five years ago. That's a pivot, a pace pivot, in the industry that came after digitization.
Paper can only move at the speed of the briefcase, right? Digitized information moves at the speed of an electron. Now, you know, we're not at Internet speed in the health and life sciences sector, but on a relative yardstick the space is moving really fast. Having been pounding away at this digitization of healthcare and introducing precision in medicine, and advocating for distributed and coordinated care, advocating for the value-based approach to care as opposed to a transaction-based approach to paying for it — Intel's been pounding on this topic for 15 years.
We're really happy to see everything moving in that direction. But I expect to see the pace continue to pick up. And the addition of training-based algorithms and AI into the mix is a giant change and a giant advance in analytic power being put into the hands of the folks responsible for the healthcare system.