Getting Data In and Out of Your Electronic Health Record

This post was originally featured on TechZulu. 

Epic is the most dominant electronic health record (EHR) company in the US. Although Epic is a private company, it boasts over 6500 employees, $2.3B in revenue, and 150M patient records.

Like most legacy healthcare IT companies, Epic runs an ancient technology stack. Its core database is a pre-SQL, cache-based database, which went out of style long before I was born. They’re currently in the midst of transitioning the front end from VB6 to .NET. Unfortunately, they decided to leave their core database technology untouched.

Epic doesn’t like working with 3rd party vendors. Epic wants to be THE solution for all hospital IT. Because of its ancient technology stack and business practices, Epic has been notoriously difficult to interface with. Epic doesn’t provide anything that even resembles an open API. They will, if a prominent customer makes enough noise, provide an HL7 ADT feed. Without getting into the technical details, I’ll say that HL7 is the opposite of a standardized, consistent API. To be fair, it’s not Epic’s fault that HL7 is lacking, it’s the government’s.

Epic recently held its annual user group meeting. They surprised the world by announcing open.epic, an open API. I almost had a heart attack out of excitement. Unfortunately, that was unwarranted. The API will allow developers to feed passively gathered wearable device data into an Epic database. The API doesn’t offer a protocol to reciprocate data exchange. Moreover, the launch was clearly a hastily thrown together soft launch. At this stage, Epic is just getting feedback from developers. They are years from a live solution at a client site.

This of course begs the question, why open an API at all? For the past 30 years, Epic has slowly built software modules for virtually every hospital function. Epic has never acquired another company, and tries to insource everything. I suspect that Epic concluded that it will never want to be in the business of manufacturing, marketing, and selling devices to consumers, so it actually found it in its own interest to open up a one-way API for passively collected patient health data.

The big winners here are Epic and analytics companies that can get their hands on Epic databases. Perhaps they’ll be able to drive new insights with wearable device data. Although there aren’t any explicit losers per se, relative to what the API should’ve and could’ve been, the big losers are the device makers that are going to feed the API. Just imagine how much more powerful this could’ve been if your Jawbone or FitBit could query an Epic database and return some intelligent insight to the patient. That’s a profound and unrealized concept.

Per the laws of capitalism, necessity is the mother of all invention. Developers want 2-way data exchange with EHRs, so they’re taking it upon themselves to make it a reality. Startups such as Human API and Validic are creating platforms that make it easy for developers to easily exchange data between previously silo-ed systems in real-time. Catalyze is simplifying the process of getting data out of complicated healthcare data files and stores.

 

Validic.png

Surprisingly, the major health information exchange (HIE) vendors, Mirth and Orion, aren’t going after these markets more seriously. They’ve been focused on connecting healthcare providers with one another,  and analytics on top of that data. They’ve been less interested in pulling in additional data from patients themselves.

The federal government has been pushing the Blue Button initiative for some time, though it’s a one-way read-only API that’s limited to claims data: demographics, problems, allergies, meds, and lab results.

The silos that have characterized health IT are being slowly breaking down, in many cases without the blessing of the monolithic EHR vendors. This is perhaps the least hyped sectors of health IT, but one of the most profound. The companies that support robust data exchange will be in a position of significant market power with strong networks effects.

Off to the races.

 

Making Sense of Calico

This post was originally featured on HIStalk. 

Google recently announced a new venture, Calico, with a stated mission of extending human life by 100 years for those born 20 years ago. Some love it. Others are a bit disturbed that Google, the company that knows (almost) everything about everyone, will be getting in bed with pharma and medical device companies.

There are virtually no details on what Calico will actually do other than analyze data to help drive medical breakthroughs. Even still, the criticisms have been loud. I, on the other hand, can’t help but think, “Why didn’t Google start this sooner?”

Let’s consider some unique Google traits:

  1. Google is the best analytics company in the world, period. The healthcare IT analytics industry is booming right now, trying to make sense of the enormous amounts of data going into EHRs. Google has been analyzing the entirety of the public web for over a decade. Google is the undisputed leader of data aggregation, management, and analytics. They do it at unprecedented, unparalleled scale. If they can use their expertise and resources to drive new healthcare breakthroughs, why not?
  2. Larry Page has stated that he wants Google to be a "moonshot factory." Google is one of the select few entities in the world with the resources and expertise to build and deliver moonshots, because by definition, they require astronomical capital investments over long periods of time. Google has publicly debuted three moonshots from its Google X labs: self-driving cars, Glass, and Loon. I find it hard to make a compelling case that self-driving cars and Loon weren’t driven out of genuinely altruistic desires, even if Google will eventually profit handsomely from them. No one except Google has been willing or able to commercialize these technologies before. Google is one of the few entities in the world investing on 10- and 20-year timelines.

The most common criticism of Calico is that Google shouldn’t be working so closely with big pharma and medical device companies. Supposedly, pharma and medical devices companies will use that data to harm the general public. That data can lead to some hyper targeted ads, but so what? Is "protecting" people from ads worth killing those who might otherwise die? Data leads to insights. Let Google crunch the numbers and figure things out that were never before possible. They’re really good at it. The cost of being wrong is… ads.

One could make the argument that it’s dangerous to let Google hoard this data. That’s kind of a moot point since Uncle Sam is already doing so via the US healthcare billing system. Google runs the best, most secure data centers on the planet. If anyone is going to house your data, it should be Google.

If and when Calico finally delivers innovations, they’re sure to be expensive. That’s standard practice. I can’t recall a single technology breakthrough in history that was adopted by the bottom 10 percent before the top 10 percent. By definition, investing for 10 years to drive breakthroughs is expensive. Personalized medicine, whenever it arrives, will be absurdly expensive. The wealthy will adopt it first. That’s fantastic. The laws of capitalism dictate that competition will crop up and drive prices down and speed innovation up. It may take an additional five or 10 years, but inevitably, prices will plummet.

If Google is successful, what will it mean for society? How will Social Security work if everyone lives 100 years longer? How about Medicare? The answer is that we’ll figure it out when the time comes. Did Henry Ford care that cars would drive pollution and global warming? Did Einstein care that his research would drive the development of the atomic bomb? Did Steve Jobs care that the NSA would spy on everyone that used a computer? We’ll adapt to the new social norms if and when the time comes.

The world will look dramatically different in 15 years, and even more so in 30 years. We’ll probably have computers in our blood streams and contact lenses, we’ll probably be printing most consumer goods in our own homes, and robots will most certainly be everywhere. We shouldn’t try to render ethical judgments on a future that we can’t possibly comprehend.

Let the chips fall where they may. If Google wants to invest truckloads of money trying to do the impossible, why should we try to stop them?

Why Google Fiber? Moat-widening Endeavor & Future of Computing.

This post was originally featured on TechZulu. 

In twenty years, Fiber will be one of Google’s most valuable assets.  Fiber is a genius long-term strategic play.

With Fiber, Google has 2 primary objectives: 1) Billionaire celebrity investor Warren Buffet explains the first reason most eloquently: “In business, I look for economic castles protected by unbreachable moats”. Fiber is a moat-widening endeavor. 2) Fiber is a means to Google’s vision of the future of computing. This post will focus on the former.

Apple, Google, Facebook, and Amazon are striving to be the super-mega technology company that owns every layer of the value chain. Google’s vision is the most ambitious of its peers. To understand that assertion, let’s observe how and when Google encroached on each layer of the technology value chain required to access any web service. The following graphics illustrate the value chain, and Google’s systematic march down the value chain. The list on the left enumerates the fundamental pieces of the value chain, and the list on the right show how Google has moved down the value chain. This is by no means an exhaustive list, but provides enough detail to illuminate Google’s long-term strategy.

Value Chain.JPG

 

Chronology.JPG

Component design – computers need computer chips in order to compute, and computer chips have to be designed by someone. So do batteries, cases, cameras, and many other components.

Component production – someone needs to manufacture all of those components.

Device design – all of the major computing components – CPU, RAM, connectivity, display, battery, etc. – need to be enclosed into a coherently designed device.

Device assembly – more than 1B cellphones are sold every year. It only takes a few million Chinese laborers and a few hundred factories to assemble them.

Operating system (OS) – computers need an OS to provide a user interface.

Web browser – even a functional computer with an OS isn’t enough to access most Google services. Computers still need web browsers to power HTML, CSS, JavaScript, and many other core web technologies.

Internet infrastructure (switches, routers, cables, cell towers, etc.) – every piece of data on the Internet is routed through dozens if not hundreds of check points. Someone needs to manufacture all of this hardware and network infrastructure to connect everything.

Internet connectivity (the ISP layer) – someone needs to pay for, deploy, and manage all of that Internet infrastructure.

Data centers – there need to be computers on the other end of the pipe to do all of that computing.

Server-side software infrastructure – virtually all modern web applications rely on virtualization, scaling, load-balancing, and database technologies, among many others.

Web applications – and finally, at the very end of the value chain are the web applications that people actually connect to, such as Search, Gmail, Maps, etc.

Google started at the very end of the value chain: search. Search is just another web application. Once Larry and Sergey developed search’s secret sauce, PageRank, they backwards-integrated into server-side software technologies to power and scale search. Although Google doesn’t create every piece of server-side infrastructure that it needs, it has developed quite a few.

As Google grew, so did its data centers that powered search. Those same data centers would later power the rest of Google’s services. Data centers are a strategic layer of the value chain: data center efficiencies and deployments are some of Google’s key competitive advantages and trade secrets. It’s extremely difficult to setup and operate data centers at Google scale.

Google developed extremely sophisticated expertise in the final three layers of the value chain before their IPO in 2004. Since then, Google has consistently encroached on the first half of the value chain. The first half of the value chain is far more capital-intensive, so they weren’t able to move into those layers until the company grew and matured. Google understands that it relies on hundreds of other companies that come before it in the value chain to deliver users to Google’s services. Google doesn’t want to allow anybody before itself in the chain to divert users away from the immensely profitable Google-verse at the end of the value chain.

Google recognized the enormous costs and scaling issues involved in providing Internet connectivity to the masses, and in manufacturing and distributing routers and hardware backbone of the Internet to the world. So they skipped over those steps of the value chain, and purchased Android in 2005 (launched in 2008) to backwards integrate to the OS layer. It’s not publicly known exactly when Google began developing Chrome and Chromium, but it must have been late 2006 or early 2007 since Chrome launched in September 2008.

In October 2011, Google purchased Motorola, which has business units dispersed through the first four layers of the value chain: component design, component production, device design, and device production. At the time of acquisition, the press primarily asserted that Google bought Motorola to help protect Android’s legal intellectual property in court. Although it wasn’t discussed very much in the media, vertical integration was a major motivator. Additionally, Google began dabbling in hardware design with the smartphone Nexus line in January 2010, and has since expanded into tablets, laptops, and Glasses. The hardware teams at Google are growing quickly.

Although Apple designs a few of its key components – namely CPUs and batteries – they don’t own or operate a single factory or assembly facility. The hardware production industry is capital and labor intensive, complicated, and faces significant human rights, regulatory, and seasonality risks. With the purchase of Motorola, Google entered both layers of the value chain that require owning and operating a factory. Google is the only member of the gang of four to move into the physical production layers of the value chain.

Google skipped two layers: producing the hardware backbone of the Internet, and providing Internet service to the masses. It’s clear that the ISP layer is far more important than the Internet backbone itself layer. Routers don’t do anything to discriminate against Google or anyone else. But the ISPs that own and operate the routers might. ISP executives have publicly spoken about violating net neutrality.

For the uninitiated, the Internet has historically operated per the principle of net neutrality: ISPs treat all data the same. Violating net neutrality would be analogous to installing Internet toll roads. ISPs can discriminate and collect royalties by installing toll roads on the highways to access specific websites, including Google. Google is concerned that ISPs may begin to price discriminate between websites, allowing ISPs to charge users at different rates for different web services. Just imagine if ISPs imposed a cost of $.02 per Google search, and provided Yahoo for free. Google would be in peril. ISPs are already throttling YouTube because video consumes so much bandwidth. In a world without net neutrality, the ISPs will gouge Google. Google wants to undercut the ISPs before they undercut Google.

It’s absolutely brilliant that Google is backwards integrating to provide Internet connectivity. Not only are they running a profitable business, they’re creating an enormous amount of good press for themselves, lighting a fire under the ISPs to increase their speeds, and acting strategically to protect their spectacularly profitable web-application services.

There’s a 2nd strategic reason to integrate to the ISP layer: super-fast Internet connectivity will power Google’s vision of the future of computing, in which we approach 100% computing in the cloud. In order for this vision to pan out, connectivity needs to be so fast that people can’t tell the difference between native and web applications. When connectivity is fast enough, it doesn’t matter which end of the pipe computes. More to come on this in part 2.

Over-Communicate

I'm an email junkie. I'm blessed that I can type 100 - 120 words per minute. I write more emails than anyone that I know. Most of them are one-way emails that don't receive a response. I'm proud of that.

Why? Because I believe in over-communication:

1) Human psychology tends towards under-communication. Under-communication leads to passive-aggressive bullshit, hence the divorce rate, unhappy employees, and all kinds of political warfare. Nothing is worse than passive-aggressive bullshit. Over communication stems passive-aggressive bullshit.

Unfortunately, people naturally believe that others know what they themselves want to be known, and that others don't know what they themselves don't want others to know. But in fact the opposite is true: it's known as the spotlight effect. Human psychology tends towards passive-aggressive bullshit.

As a CEO, I'm a professional beggar, storyteller, and fortune teller. I spend all day talking to others either verbally or in writing. I don't do anything except communicate.

2) I want to be known as an over communicator. I ask people for all kinds of stuff: money, access to people that have money, people to work for me, if they know people that might want to work for me, etc. Under-communicators don't realize that many times, they are only in communication with someone to ask for something. Nearly 100% of their conversations are a one-way ask. People don't respond well to that.

Instead, as an over communicator, I'm always offering help, always trying to keep people updated, and always looking for feedback. When I make an ask, it may only be 20% of our total communications. People perceive you dramatically differently when you've been genuinely engaging before you ask for help. The difference is remarkable.