The future of a truly connected, more efficient and effective, healthcare system, one that is a true learning system, is eminently possible, and the technology is available to make it happen—but moving forward will require healthcare leaders to act on certain priorities, Eric Schmidt told a capacity audience of healthcare IT leaders on Monday afternoon. Schmidt spoke to HIT leaders in the 8,770-person-capacity Palazzo Ballroom at the Sands Convention Center on Monday afternoon, during his opening keynote address, at HIMSS18, the annual conference of the Chicago-based Healthcare Information and Management Systems Society, which, it was announced, had drawn more than 43,000 attendees to the conference.
Schmidt, best known as the executive chairman of Google from 2001 to 2017, and of Alphabet Inc. (Google’s parent company) from 2015 until December 2017, outlined his vision of that future for his audience of healthcare IT leaders. He began by asking audience members to imagine a future in which a physician would engage in a patient visit, assisted by a form of technology--a virtual assistant--that would “listen to the conversation, provide [clinical decision support] advice in his or her ear, and fill out the transcript [the clinical documentation of the visit] for the doctor.” That technology would be unobtrusive, but would relieve physicians of all the hours of documentation they do every day, while also providing the data needed to fuel population health management and other key goals of healthcare leaders.
“This technology—everything I just described—is buildable today or in the next few years,” Schmidt told his audience. “All it takes is all of us, everyone in this room—to figure out how to build it. I’m going to give you a roadmap,” he added. “I’m going to start with, get to the cloud, run to the cloud. Take an airplane, fly to the cloud. Most of you sit in data centers that work on proprietary logic. We now have cloud technology available, from Google and others, that’s much safer than your data center, much more compliant than your data center.” Why aren’t patient care organizations moving more rapidly to the cloud? He asked. “The cloud is more secure. And I don’t want you repeating the infrastructure work we’re doing, but rather to focus instead on the innovation.”
Schmidt continued, “At the same time, a revolution is taking place in my industry. Scale changes the rules, changes everything.” The key building blocks of the healthcare computing of the future? “The cloud. Neural networks and reinforcement learning. The explosion of networks. If you take data, and feed it into auto machine learning, it will automatically feed you information,” he said. “That’s how fast this is working. In terms of machine learning, which is what the primary progress in the next few years will be about. Where does it apply?” Diagnostics, genomics, and “medical imaging, which is largely a solved problem.” And when healthcare leaders put together the following elements—“data plus cloud, plus powerful networks, plus deep learning and reinforcement learning”—that combination of technologies will fuel advances in diagnosis and treatment, as well as population health management, that sound futuristic now, but that will change healthcare within a decade, he predicted.
Google’s announcement around healthcare and cloud
Of course, Schmidt’s speech was not given in a vacuum. Indeed, Monday morning, in a blog posted Monday morning, Gregory J. Moore, M.D., Ph.D., vice president of healthcare at Google Cloud, wrote, “Google Cloud’s goal for healthcare is very much a reflection of Google’s overall mission: to organize the world’s information and make it universally accessible and useful. Applying this mission to healthcare means using open standards to help enable data sharing and interactive collaboration, while also providing a secure platform. Just imagine if all healthcare providers could easily, securely and instantaneously collaborate while caring for you. Ultimately, we hope that better flow of data will inspire new discoveries with artificial intelligence (AI) and machine learning (ML), leading to insights that improve patient outcomes. This week at HIMSS,” Moore wrote, “we’re showcasing our progress toward serving this mission through our Google Cloud Platform (GCP), G Suite and Chrome solutions, our work with customers and partners, and our focus on compliance and security. We’ve recently launched the new Cloud Healthcare API, which addresses the significant interoperability challenges in healthcare data. The new API provides a robust, scalable infrastructure solution to ingest and manage key healthcare data types—including HL7, FHIR and DICOM—and lets our customers use that data for analytics and machine learning in the cloud,” Moore wrote.
Meanwhile, Schmidt told his audience on Monday afternoon, “The really powerful stuff, right at the edge of what I do, is prediction. It’s one thing to be able to classify; it’s another thing to predict the next step in an outcome. That’s what we want: it allows clinicians to predict things. We could predict outcomes in the ER, for example, up to 18 to 24 hours in advance of the current systems, because of the deep analysis available. Shakespeare said, ‘We defy auguries, that we can’t predict our own fates.’ But machines can,” he argued. “As I age, I want the computer to make sure I have a long and healthy life.”
Schmidt went on to describe a whole list of potential applications of the new technology, as well as to describe some areas in which some advances are already foreshadowing future potential. “There are areas where we’re getting some wins,” he said. “Take for example sudden cardiac death, a huge killer worldwide. Almost all of them seem to be predictable and are probably delayable. If we can move one step faster, we can get action faster. ECGs. The current ECG approaches and tools are thought by physicians not to be reliable; that’s an obvious case use for machine learning. Atrial fibrillation can lead to stroke risk. Who should be on blood clotters? Algorithms for diagnosis have been developed over years” in that area, he said. “But if we use machine learning and do historical analysis, the benefit in terms of lives saved is probably the equivalent to a new drug and far, far cheaper. And the retina is a view into your vascular system of enormous value. We just published a paper,” he added, “where we took retinal images and accurately predicted heart risk factors better than doctors, better than a1c, heart rate, other factors. So imagine if they were to take retinal images and train doctors to identify the path of the disease, and they will know exactly when to intervene. That model already exists for diabetic retinopathy and macular degeneration, but the power will be in cardiac care.”
What’s preventing progress?
Addressing the inevitable question of obstacles, Schmidt asked, “Why is this not happening faster? There are many potential reasons. We’ve got the technology, it could add value. But there’s something missing. There’s the lack of a ‘killer app’—of something that causes all the data and information to become rationalized. People forget that Windows 95 did not initially have Internet on it; that was seen as an add-on. And before the smartphone came out, which was a killer app, phones’ data wasn’t integrated.” So the Internet itself, and email, and smartphones, were all examples of the “killer app” to which he referred—transformational technologies that changed the landscape of computing forever.
As for the challenges facing healthcare before this industry fully enters the future of computing, Schmidt said, “The point is that this transition needs two killer apps. The transition will go from workflow-based to voice-based. And we want everybody practicing up to their level of license, with the patient involved. So what does it take to make Liz?” he asked, referring to the name he gave the concept of the mechanism that could better support physicians in patient visits. “A common digital health data store; normalized data form all sources; longitudinal trend analysis, and the ability to predict next steps. You need voice transcription, voice translation, and multi speaker disambiguation, and language recognition. We’re much closer to the vision I outlined. This is fundamentally a search problem, and Google is very good at search problems. But we need everyone to work together.”
Indeed, Schmidt conceded, speaking of the foundational work that will need to be done to bring all the technologies together and advance healthcare into the future, “This is really hard, it’s really humbling, and it’s complicated. But if we all work together, we can really save lives at a level that’s unimaginable. I predict that in ten years, if you invite me back, I will bring the equivalent of Liz [the physician technological facilitator solution] to join me on the stage.”
After he had concluded his formal speech, Schmidt sat down and had a conversation onstage with Hal Wolff, the CEO of HIMSS. During that conversation, Wolf noted that “Artificial intelligence has already been built into many applications. Can you name a few examples that you might see in the exhibit hall here at the conference?”
“There are many,” Schmidt responded. “The easiest ones to use are the various speech integrations that Google, Apple, Siri, do; Google translate, that’s all machine learning. These things don’t have dictionaries or databases; it’s all learned. Machines can discover structure just by seeing enough strings of Spanish or Russian. In fact, the translation tools don’t make use of dictionaries, only patterns. And they can impersonate the sounds of language.”
“How do we accelerate [progress]? Is there something we can do to help in that preparation?” Wolf asked.
“This stuff’s happening too slowly, because you’ve not moved to the kind of platforms that scaled my side of the industry—complete interconnection, data architectures that are open, the cloud,” Schmidt said. “My view of the sum of the IT industry is that it’s conservative beyond where it should be. You care about security? The cloud is going to be more secure than your data architecture. You’re worried about your data being stolen by the bad guys? It’ll be safer in the cloud.”
Prior to Schmidt’s speech, Wolf had given a brief speech to the audience, and had emphasized that fundamental change was already making itself felt in healthcare. Until recently, he noted, the healthcare industry has suffered from “the lack of actionable information. This is a critical moment for healthcare,” he said, “because we have really come to the point where we’re really beginning to take advantage of our investment in IT and in the critical infrastructures we’ve been building. We’re beginning to use the massive amount of data that comes out every day. It’s not just information that comes from within, but also from without,” he added.
“How we harness the information and begin to use it, this is our next critical step,” Wolf said. “And in support of that, we brought forward the concept of the power of ‘and.’ For years, our mission statement at HIMSS has been ‘better health through IT.’ This year, we made a change, and the board looked up and said, ‘better health through information and technology.’ Our infrastructures are going to be there—they are our bedrock and foundation… but information will be where we go and drive, in terms of AI, machine learning, clinical decision support, clinical pharmacy, how we use data from a variety of sources. It will be the next level of delivery” of data and information.
The conference continues through Friday, March 9.