Putting It All Together

I've just spent two stimulating days in DC, first going to the HHS/IOM Health Data Initiative meeting (aka "Datapalooza") and then at an HHS/Kaiser-sponsored Health Innovation Summit (see #futurehealth on Twitter).  What's clear is that this is an exciting time for innovation:  we're seeing companies pledging to enable patient downloads of their data (I heard "bluebutton" used as a verb for the first time); more releases of federal population-level data; and a gaggle of companies leveraging these data to offer terrific wellness apps.  The future's bright, indeed.

In thinking about how the pieces come together, the challenge seems a bit different now.  A few years ago, a key question was how to leverage the data in one's own personal health record to build apps that would help people take care of themselves.  An insight of the Pioneer-funded Project HealthDesign was that the data needed to drive those apps came not so much from the medical record but rather from the flow of life: observations of daily living (ODLs) about diet, exercise, sleep, mood, pain and more.  So apps needed a platform that would integrate both medical record data and this new set of patient-generated ODLs. 

Now, as I look around, I see multiple categories of data that one might want somehow integrated:  in addition to medical record data and ODLs, there's the population-level data that HHS and others are releasing (that can provide useful context and benchmarking), environmental data (mashing up your own geo-location data with data sets about the environmental factors associated with those locations), genomic data, and, of course, if you're the competitive type, comparative data from your social network.

What I've observed is many siloed apps.  I've got devices and web sites that either capture data (like the treadmill to the iPod to Nike+) or give me the opportunity to enter data (like putting my weight into Google Health), but nowhere that stores it all.  That doesn't bother me so much because at one level, where it's located isn't all that important as long as I'm confident that it's secure and reliable.  Missing are the abilities to a) present all of these disparately gathered data in some context that gives them meaning and b) to analyze these data to give me insights about patterns and correlations that would help me understand how to be healthier.  Where are the apps that will crawl across the different data stores and pull it all together? And what will they need to work? 

I'm curious to hear of companies that are trying to make this happen.  What are the best examples people have seen?

 

Comments

[Jane Sarasohn-Kahn]

This is the real heavy-lifting: meaningful, real Connected Health can't happen until there's a 360-degree view of our personal health data, marrying what our health plans know about us (residing in claims databases, where so much of our personal health data live in the grandest silo of all in today's U.S. health data ecosystem) to all the other data silos -- including our ODLs generated through apps, USB-powered blood pressure meters and digital glucometers, FitBits, and Withings scales. Integrating the data from these sources is Job 1. Job 2, then, will be the ability to bring the right metrics together for the question at-hand: for example, "What can I do to improve my sleep?" or "I'm exercising hard and eating right; why am I not losing weight?" Getting the right data into a personal health dashboard (for lack of a better word) will require more than an app: this is very complex data integration and interoperability. Can my doctor's Epic EHR now accept my Withings and FitBit data? Not yet, but he knows I'm keen to have this as part of the patient portal. This would represent Meaningful Use to me. Judith Faulknet EPIC CEO, are you listening?

[Steve Downs]

Jane,

Thanks for the comment. I agree -- heavy lifting, indeed. I like your point about your patient portal. The data wouldn't necessarily have to reside in Epic for them to provide you with the data in your portal though -- they (and other EHR vendors with patient portals) could take a page from the personal finance industry and serve up both data that they maintain along with data scraped from sources that you point them to. I'm less concerned with the challenge of integrated display -- I feel that's been solved in other contexts -- but having the ability to analyze across the data for the insights you describe does seem a lot harder.

[Paulo Machado]

Excellent post & comment!
We are moving in the right direction.

Acceleration of the process will occur when
1-actionable insight from my Personal Health Dashboard is linked to my sustained behavior changes that led to better health outcomes & lower costs and
2-Incentives are aligned across stakeholders
3-Information flows freely across stakeholders

The future is bright. The journey is long and difficult!

[Matthew Holt]

No Jane, Judy is not listening, but the data utility layer is expanding. Runkeeper (among others) is starting to accept data from lots of other devices and the same distinction between data and applications is happening between storage and metrics in the device world. Soon it will all flow more easily into one place...and Healthvault (if not Google) is continuing to push the ability to move multiple types of data around and do things to it. Finally intelligent network based permission brokers are helping (Resilient Networks being one).

We're not there yet, heck we may never be "there" but we're getting closer

[John Moore]

Hi Steve et.al.,
Some good comments here which by and large i agree with. However, what I do not see being addressed is how to release the data - as Todd likes to call data liberation! It is still too difficult to get data/medical records, in a computable digital format from your local doc, specialist, hospital and even when you do, they oft-times withhold certain data types (eg lab values). Payers are not much better with one's claims data - although one could argue as to the relative value of such data.

So first, we need to get the data following, easily, and I can not emphasize "easily" enough. I find it challenging at times and I work in this sector, this is part of my job so imagine what it is like for the average consumer? It is still far too difficult. Makes one wonder if this is done intentionally or simply a lack of desire on the part of those holding the data.

But once the data flows, we need to aggregate, shake and bake it to produce something meaningful for the individual, to help guide them, their lifestyle, their behavior(s) to improve their health and well being. A long path ahead to making that a reality.

Though I try to stay hopeful and optimistic it is indeed a challenge in light of the challenges we face to achieve that objective of "Putting it All Together"

[Steve Downs]

Thanks all for the comments. John -- that's a helpful reminder that momentum in the direction of data liberation is not the same as widespread liberation itself -- there's a lot of work to be done and a lot of pressure to be maintained. And Matthew, I like the notion that applications that started out narrow will expand to integrate with others and can conceivably evolve as competing platforms. Thanks for the tip on RunKeeper -- their Health Graph (see http://blog.runkeeper.com/new-feature/health-graph) looks like a good move in this direction.

[Gary Wolf]

Very good discussion here - thank you Steve. I suspect that the vision of free, open, well-integrated health data is going to function as an ongoing orientation and strategy, rather than an end-state, just as the larger notion of open data and free collaboration guide developers, users, and critics guides software in general. (How could it be otherwise, given the insights from ODL research, Quantified Self, etc.? If health data is best understood as part of a larger ecosystem of personal data, then all the tensions of platform&software development will play out in the health space also.) For instance, integration can be a valuable function of walled gardens, where the "owners" of the most specialized data, such as physician records, "invite" users to contribute their ODLs. This comes at a cost to openness, of course. But in an "open" system built around APIs, integration and ownership can occur at the level of interpretation. So, for instance, platforms that display data meaningfully for users and use proprietary algorithms to interpret data that is collected via APIs, can become effectively closed. We're watching this process with search, advertising, email, calendars, and of course "social media." Since it is impossible to predict all the ways that our data will be meaningful, we can't know in advance how conflicts over openness and integration will play out concretely. Instead, the conversation we're all having helps establish a kind of ethic of openness and integration that comes in handy when we make specific requests (demands) of the systems we use. So for instance, when Hugh Campus asks "why can't I see the data from my own ICD" he's doing more than seeking a personal benefit. He's helping establish a general sense of our rights.

[Thomas Goetz]

Great topic, Steve. A few comments, drawing on my own attendance at both DC events. I find it compelling that personal health data is now being created/liberated in two distinct areas: one is the QS/patientslikeme/self-tracking stuff, which is new data that wasn't being captured previously. As you and Gary note, the more openness here the better, and though the pace isn't moving as fast as we'd like, there is some promising API movement.
The other area, though, is extant health data (mostly via EMR/EHRs). Here, liberation is just the first step: we must also create tools to actually *do* something with the data. This, in many ways, is a harder area to tackle, given everything from HIPAA to institutional/practioner resistance (as John Moore notes above). But it is, I increasingly believe, a potentially more powerful data-set, both to the individual and to research.
I found it very encouraging to see things like Blue Button emerging as central standards outside of government (Aetna, for instance), and I found it even more encouraging just to see that there are interfaces between two communities that, until recently, had virtually no shared language: the policy/DC/health establishment and the entrepreneurial technology-centric startup crowd.
The fact that there is now an interplay, that a shared language (around patient engagement and improved outcome metrics, most explicitly) is being codified, is reassuring to me. As this language develops, I think the dots are starting to connect, and as a network of functioning information emerges, the advantages to participate start to outweigh the advantages to remain proprietary.

[Steve Downs]

Gary and Thomas -- thanks for the comments. The coming together of the two worlds is very interesting. Very positive, I agree, as they have been far too separate until recently. (Thanks Todd Park and also Matthew's done a lot in this regard). They are also very different worlds, with different practices, different industry structures, different IT paradigms, different laws and different cultures. How they come together is part of what we're exploring in Project HealthDesign (www.projecthealthdesign.org) which is attempting to bring ODL data into the clinical encounter (and, perhaps more importantly, patient-clinician communication that can take place between the encounters). We're finding a series of challenges related to the difficulty of adding new data to EHRs, the potential liability associated with having too much patient data, and the ridiculously busy day of a practicing physician. One hopes that these are transitional issues and not deal-breakers. And Thomas, your comments make me more optimistic as a shared language and perhaps at some point a shared vision could start to make these challenges surmountable.

[Deborah Estrin]

Steve, I agree you are focusing on exactly the right point here...the sense making. There are good examples in other (and yes, far easier) domains such as Mint.com and Tripit where services scrape together and make sense of disparate data sources and formats for each individual client. We can learn from these but i think we all would agree that the public will be best served if we take a much harder look at supporting patient-owned data, patient-controlled selective sharing, and built-in transparency tools so that data provenance and flow is audited and legible. Moreover, in this early stage of mHealth development I hope we can promote an open ecosystem of tools, techniques and data to create a rich evidence base and promote shared learning.

[Steve Downs]

Thanks Deborah -- and people can follow your work in promoting that open ecosystem at www.openmhealth.org.

And nice article in Technology Review today!
http://www.technologyreview.com/biomedicine/37870/?ref=rss