Crafting Health Care’s Future at Kaiser Permanente

Deep in the heart of an industrial park in San Leandro, California, is an unassuming warehouse that could house anything — consumable goods to pipe fittings. But open the doors and you find a technologically advanced living laboratory — the Garfield Innovation Center — which is part of Kaiser Permanente’s integrated health care delivery system.

The center brings together collaborative teams from across health care disciplines, including engineers, architects, technologists, physicians and nurses. Together, they’re piloting innovative care solutions and developing a vision for what U.S. health care might look like in the future.

Dr. Yan Chow is a director in the Innovation and Advanced Technology group at Kaiser Permanente. A physician with over two decades of primary care clinical experience, Dr. Chow also has a keen interest in technology. He has founded several technology startups and is a software and database developer. He is co-inventor of ultrafast network storage technology with three U.S. patents. His areas of expertise include health care IT innovation, telehealth, big data and analytics.

Dr. Chow spoke with MIT Sloan Management Review contributing editor Renee Boucher Ferguson about his work at the Garfield Innovation Center and the impact of data and analytics on the future of health care.

What is the role of the Innovation and Advanced Technology group within Kaiser Permanente?

One of the reasons that Kaiser Permanente seems to attract a lot of attention from technology startups is that many of them have models that cross silos. If they were to do any kind of a proof of concept in the health care market outside of Kaiser Permanente, it would be prohibitive. Inside Kaiser Permanente, because of our integrated care delivery system, we have access to pharmacy data, lab data and all kinds of other data.

I joined the Innovation and Advanced Technology group as a founding member in 2006. At that time, there were three visionary leaders at Kaiser Permanente. One was Christine Malcolm, who was heading the National Facilities Services. That’s the group that designs and builds hospitals and clinics for Kaiser Permanente. There was also Dave Watson, who was the chief technology officer in IT, and also Marilyn Chow, who was the chief nursing officer.

They decided to create two structures to assess the future. They felt that, strategically, technology was going to be a big, big lever in the marketplace, and even though we had adopted early in terms of electronic medical records, we still needed to know what else was going on outside Kaiser Permanente.

And so they set up the Innovation and Advanced Technology group and the Garfield Innovation Center. It has essentially become a world-famous facility for care-delivery simulation.

Can you describe the Garfield Innovation Center?

It’s a 37,000-square-foot warehouse, repurposed. It has template designs for future hospitals. It has an entire med-surg hospital ward, built inside, with patient rooms of the future. There’s an OR [operating room] of the future, labor and delivery of the future and newborn ICU setup. There’s a home environment to test telehealth. There are micro clinics and all kinds of technologies that go into the center on a rotating basis so we can imagine how we might practice health care in five, 10, 20 years and so on. That’s the physical facility.

Our group was formed in conjunction with the Garfield Innovation Center as the think-tank component. Our mandate is to go out and look for new and emerging technologies that we think will be coming into health care over the next two, five, 10 years. Our main focus is in three areas: Telehealth/telemedicine, mobile health, and analytics, especially in the last two to three years.

We also focus on things that augment reality — virtual reality — and on biometric authentication, social media, gamification, genomics and all kinds of areas that would impact our IT infrastructure.

When you say “focus on,” does that mean testing technologies and implementing them at Kaiser Permanente?

Yes. We have looked at over 2,000 technologies in the last eight years, and did deep dives into between 300 and 400 of those. About ten have gone into production at Kaiser Permanente. This is not surprising, considering that most of the stuff we look at is pretty far out, in that it isn’t really ready for prime time. A lot of startups have stuff held together by duct tape. It’s very interesting stuff, though, and it’s actually a very interesting time in health care right now. That’s on the outside — solutions looking for problems.

On the inside, about six years ago, our CIO Phil Fasano launched a program called the Innovation Fund for Technology. It offers all 175,000 Kaiser Permanente employees the chance to apply for funding for a pilot project based on a good idea. We’ve had over 700 applications and funded 84 projects at various levels. Of those, about 18 or more have gone into production at Kaiser Permanente.

In the last three or four years, we’ve also been very active with innovation exchanges with other innovative organizations, both inside and outside health care. We’ve met with Mayo Clinic, Harvard, MIT, UPMC, Stanford and so on in health care. Outside health care, we’ve met with Wells Fargo, Safeway, Comcast and a lot of consumer electronics companies, because even they are seeing the value of building health care type of capabilities into their products.

In the last two to three years, we’ve also been building relationships with investors and venture firms that have brought us their portfolio companies to look at. I think this all points to the real need in health care for reality testing.

Among those different avenues that you look at — the external and the internal, the startups, the employee grants — where does analytics come in, and where do you see the future of analytics for Kaiser Permanente?

Analytics was identified as a key focus, a strategic area for us, because Kaiser Permanente is where other people will be in a couple years.

We’re the leading edge because we adopted electronic medical records early. And we are coming to the place where we’ve collected so much data — probably over two terabytes a day of data — on patients, that we’re now asking the question: what is the value of the data? And that’s a very natural question. Besides the regulatory reporting and the meaningful use, what is the value of the data for improving outcomes and improving affordability? The mainstream will probably hit that at about two to three years. Once they finish installing all the medical records under the HITECH Act, and they’ve become comfortable with them, they’re going to ask the same question.

Being ahead of the curve has presented a challenge, because the technology isn’t there yet. Both technology and the legal, regulatory, standard of practice piece — all those kinds of things that go along with technology — are also lagging. But we believe it’s very strategic, because we have all this data.

The other thing that’s really interesting about Kaiser Permanente is that our integrated-care delivery system results in patients who stay with us for a long time. In northern California, for instance, our members on average stay with us for 17 years. So that’s a very long longitudinal database. Whereas in the general market, the average tenure that people have with their health plan is about two to three years. When they leave and go to another health plan, that data continuity is lost. There may be a different vocabulary, a different coding system between the two health plans, so it’s just really hard to maintain the data. I think the move towards ACOs [accountable care organizations] will help with that, but it’s going to take a while. So, having that integrated, long-term data is really valuable.

What have you been doing with all of that patient data?

Medicine is evidence-based. We’ve been looking, from a research point of view, at the data forever. For at least two to three decades we’ve been looking at our de-identified data from a population/disease management point of view. And we’ve been able to generate a lot of research, as well as best practices and best-practice recommendations, based on that huge database. So that’s been good for helping the general state of medical arts, to help people practice better medicine.

But ultimately the goal is to personalize medicine, which means we have to have a much more granular view of patients, so that we can intervene in patients’ conditions with a very specific, personalized intervention that’s the most efficient — most cost-efficient, most acceptable, and most effective — much more so than today. So, how do we personalize? We can’t use randomized clinical trials, because they don’t fit every patient. They’re not personalized. And also, they take too long, way too long.

So now, the move is towards statistics-based analytics, the correlation of analytics. That’s the only hope we have to keep up with changing data. That move to a data-driven practice of medicine as opposed to evidence-based medicine, which is causal — data-driven is statistical, correlational — that’s a big shift.

If you think about it, that’s a quantum shift in the way that we practice medicine, because it has a lot of ramifications that I don’t think people have really thought through. So for instance, if a doctor says “well, based on 300,000 people like you, and also your own particular behavior and your own particular side effects, this is the intervention that we recommend,” then the question is, suppose something goes wrong? You don’t have randomized clinical trials. You don’t have any causal information.

That’s a big issue. We have to have the medical peer organizations, the professional organizations, actually decide yes, that’s a good way to practice medicine, because that’s the only way to keep up with the data that’s changing all the time.

And so if you look at it like that, then it’s like two ends of the spectrum. On one side is the gold standard, which is randomized clinical trials. On the other side of the spectrum is anecdotal medicine. That is, the practitioner who’s practicing in his own practice, her own practice, and has an experience over their career with certain kinds of patients. And based on a limited statistical study, statistical experience, they treat a certain way because it works. And so that’s anecdotal medicine. It’s not causal, it’s just basically experiential.

What we’re doing today with analytics is we’re boosting that end of the spectrum by like a million-fold. Instead of an experience of 30 patients in a career, there’s an experiential 300,000 patients over a huge population. And that’s the promise. It’s sort of like creating anecdotal medicine superstars. So, it’s creating the ability to use anecdotal, experience-based medicine on a large scale.

Looking at the bigger picture, you’ve got all of these implications and issues to solve with data. Where do you start? Or where are you on a time line of already having started?

We actually have some things in production. For instance, in southern California, we have a very sophisticated natural language processing group that was recruited out of Stanford. And they are actually in operation. That is, they’re analyzing revenue data and claims data to capture lost opportunities in terms of coding. That’s been actually very successful. But that’s the only instance I know of, of real advanced analytics being done in the operations. The rest has been on a research basis.

Like I say, we have an innovation fund program, and I believe we’ve had maybe two or three projects doing NOP stuff. There’s been some proposed projects for analytics for workflow in a same-day surgery unit and things like that, that’s very isolated. So the experiences are scattered, but people are just starting to think about it. And obviously the area is not new. People have always thought about how we can optimize workflow, how can we predict who’s going to be readmitted to the hospital in 30 days — all those big questions. And we try to do that manually, but now we have access to new kinds of engines. So, it’s a very exciting time.

But the vast majority of clinicians just want to do their work. The people who are thinking about it are the few innovative physicians or innovative IT folks or health-plan folks that can see a lot more potential in what we’re doing.