The healthcare industry has been waiting with baited breath to see what nontraditional players in the space do next, with some worried that high profile entrants like Amazon and Google will dramatically shake up the industry and leave little room for them.
Google seems more interested in partnerships, however, with the past year especially bringing a host of new alliances. Nowhere is this more obvious than with Google Cloud — the company's initiative to wrangle the reams of healthcare data into storable, usable forms.
To date: The National Institutes of Health is using Google Cloud for biomedical research in its STRIDES Initiative; Google Cloud recently joined a consortium to identify and combat sepsis in hospital ICUs; representatives from Google are joining standards body HL7 (in charge of FHIR); Google Cloud is collaborating with nonprofit research center MITRE for a synthetic dataset of all residents of the state of Massachusetts — and that's not counting Google's tech partners within healthcare, which include Novo Nordisk, Flywheel, Life Image, Imprivata and more.
Healthcare Dive sat down with Google VP of cloud healthcare and life sciences Greg Moore at HIMSS last week to talk about the giant's suite of services, the benefits of the cloud and Google Healthcare's API on its one-year anniversary.
This interview has been lightly edited for clarity and brevity.
HEALTHCARE DIVE: What is Google excited about in the healthcare data space right now?
GREG MOORE: What's been happening, and what I think everyone has seen, is the amount of data we generate in healthcare is rapidly increasing. It's doubling about every 73 days now, if you include genomics and imaging data and the exhaustive data including all health and wearables and IT. At Google Cloud, we're a data company — whatever companies say is data. And in healthcare and life sciences, our mission isn't too much different from Google's overall mission. It's to organize all of that data from customers and partners, it's to make it secure and compliant, so whatever we can do to make that data compliant in a way that is incredibly secure.
We went from an ecosystem where people were wondering if data was secure in the cloud to now, if your data isn't in the cloud, it's less secure. So security and compliance is core of what we do and one of the things that Google excels at. Then we want to make that data accessible. For us (and for me as a physician) that's something I'm passionate about. That means making the data open and accessible, so that's this whole push with interoperability. So we've designed tools and services, our entire platform, that's designed to be interoperable. We've been very vocal about this, even on the national scene. Finally, it's the useful part, which is often the most exciting. That's analytics and machine learning. This is a core differentiator for Google. The incredible engineering talent and investment that we've made as a company in AI, having eight products with over a billion users all powered by AI — it's just a real core competency and it's exciting now to bring that to the market.
What is Google doing right now in the sector?
MOORE: We have an incredibly secure, compliant platform in which we've developed tools unique to the healthcare industry. The Google Cloud Healthcare API has incredible momentum in the industry to really bring these silos of data together — it's something we launched at HIMSS last year. That API — and again, this was a year ago — enables FHIR, it enables cloud technologies that speak healthcare in their native languages. And this large amount of data coming in really comes in in data silos. So there's electronic health records, imaging systems we call PACS, and genomics or lab systems: all different silos of information you could use an API approach and interoperability standards to ingest, normalize and find useful things for them, to put it in a practical sense.
Any healthcare organization would have a hard time answering this following query and, as a doc, these are queries that I'm passionate about: think of yourself being a breast oncologist that says, tell me the number of women in my healthcare system that are over age 50, that are BRCA1 positive (the breast cancer gene), that haven't had a screening mammogram in the last two years — you would think that's a simple query. But that query touches on the electronic health record in terms of the population demographics, BRCA1 positive touches lab and genomic databases and mammograms over the last two years touches the radiologists and the PAC systems. These systems don't communicate. So what this typically looks like for this single simple query — you can think of hundreds that a large health system would want to make in any given day — takes an analyst to go away for a couple weeks, bring us a spreadsheet, come back with something that's outdated and mostly, maybe 90% accurate. But not really. And that's state of the art.
One of the reasons I came to Google is we need to fix this. This isn't acceptable. That's why I've started and why Google's started to say 'hey, how can we break down these silos in data so providers can actually use that to help the patients.' And so that's a lot of hard work in terms of teaching the cloud, to understand healthcare data in the native data types and then put it under a common platforms and normalizing that data and then putting it into a platform for visualization.
Is there a timeline for general release?
MOORE: We'd love to tell you that but we can't let you know ahead of time. But I will say — I don't think you'll have to wait long. I can tell you that.
What type of clients does Google Cloud work with?
MOORE: We call them the five Ps here at Google — we get a little cute with that. So what's under healthcare and life sciences are certainly providers, that's an ecosystem here, payers, pharma and life sciences is another P, so those are the big pharmaceutical companies, the life sciences research. The last two we get a little cute with just because I wanted five. So product, the healthcare IT system, those medical device folks, basically anybody you see on the [HIMSS exhibition floor] that's selling a product that's either something you purchase or an IT product out there. The last P is professors. So it's the whole research enterprise. We do a lot of work with particularly leading solutions for genomics, we have announced a lot of partners doing major parts of their genomic research leveraging our cloud platforms to do that. So that's really where we first started and now we've come into connecting all these different data types. To make a difference in healthcare you need to bring all these different pieces together.
Why should any one of those healthcare organizations move its entire operation to the cloud?
MOORE: When I go talk on one of my CEO hospital callings, I make them aware of what's happening and what type of business we're doing so I can understand at a deep level where their pain points are, at then explore with them what tools we have, how they can join the innovation curve that will get them on their digital transformation going forward. So we take a deep partnering approach, whether it's a pharmaceutical company or a major healthcare system provider to say, what are your points and how can we take the tools that we have to help address those together.
The three largest reasons that I see the cloud could make a difference in that is at the CEO, CFO level and at the CMO level. We're still in this era of security. So what keeps boards up at night, what keeps the CEO up at night, what keeps the CIO up at night is data security. Over half of healthcare systems have been breached the past couple of years, and the cost of those breaches to the healthcare systems that are operating on margins of, at best, 2% in many cases — it can literally wipe out healthcare system profitability and margins and mean that they can't staff a nurse in the ICU to take care of a loved one.
So, how can I predict the privacy and security of my patients with this really sensitive data and to do so in a way that's cost-effective and in a way that I probably can't afford to do myself, given the sophistication level? So people are coming to us still, even a year ago when we said please come to the cloud — it's in everybody's best interest for the U.S.'s healthcare database to be secure. But people are still saying, 'Can you help me with my security, I've been breached.'
What's holding people back from making that transition?
MOORE: I think healthcare is an industry that's been really conservative and slow. It's still an industry where 85% of transactions happen over fax. It is about creating education and talent to manage things in the cloud. It's bringing people up to speed on education, getting them comfortable with the technology. And it's amazing how that's happening now but that will accelerate coming forward. So I think the real desire now is the cloud. It is about enabling that.
The other very large reason I see people coming to the cloud is they really need insight from the data they're producing, either for business purposes or for clinical purposes. So that's using something we call BigQuery, it's our big database, our beating heart engine of analytics in the cloud. It's a very powerful analytics engine where we bring this data to. So that allows users to actually organize that data and begin to have insights from it.
There's a critical need, whether that's on the business side of their operations, how many no shows can I expect today, what's my operating room schedule, what's my staffing look like — these are base tools for running a business. We've begun to help clients use these tools to say, 'are patients are getting really sick in the hospital that we're not aware of.' We're enabling [Emory Healthcare] to use cloud to fight sepsis in that initiative. And that's really compelling and it's not actually on the business but actually on the clinical side, we're enabling large healthcare organizations to give them the tools in cloud to be able to get insights on their patients to help them in their care.
What's really important, why they're coming, is that we've in essence democratized machine learning and AI. Normally you need a cadre of data scientists and machine learning engineers to actually allow them to use these tools. This is a differentiating factor [between Google and its competitors]. So you can bring data and labels only and use something called auto machine learning — no coding at all, you bring a dataset with labels, you upload it and Google will do the algorithm and bring it back to you in an API. You need no coding experience at all to do that. This uses the horsepower of all of Google's algorithms and tensor processing units to actually compute the best algorithm and bring that back for your own use. We don't use that data for our own purposes — this is yours.
We have differentiating tools like de-identification of healthcare data which allows analysts and researchers to take that and be able to do that with a button press. We have tools to allow healthcare systems that have very expensive data warehouse, they've invested tens, sometimes hundreds of millions of dollars in those, the servers they have in there are sort of obsolete in six months to 18 months, it's a huge sunk cost. We take that to the cloud, you pay by the second for what you use, you're never paying for building, or the infrastructure. And so that's really healthy for the bottom line, when people take care of the data. And with the way that healthcare data's growing right now it's sort of a nonstarter for a CIO to go to their chief financial officer and say hey, I need to build another data warehouse and by the way, I need to put another one on the budget for six months from now. That's just not sustainable. They have to keep the data but they're not getting value from it either.