With all the hype surrounding big data, pinning down its ideal usage is important for planning the development and expanding adoption of this technology. What goal should the healthcare industry have in mind as it explores the possibilities for improved care and lowered costs that big data presents?
"People that talk about big data often talk about...creating an open source network that somehow aggregates all this data and we can get better outcomes," Zach Landman, MD, and CMO at DoctorBase, told Healthcare IT News.
Landman went on to draw an analogy to Facebook. He suggested that "more and more people over time could potentially be open to the idea of sharing their information for the greater good," although he noted that it is unlikely that patients would agree to share their patient data in a public manner similar to how people share personal information on the popular social network.
Still, without even beginning to discuss - though in no way belittling - the serious privacy concerns such a health information network would raise, another issue obstructs current efforts to make the most of big data.
Big data, true to its name, comprises massive amounts of data, but what sets it apart from traditional data warehouses is the variety of data sets from which analysts try to derive insights into how to better target care in the population to which the data refers.
Locked in the system
The issue, as Landman put it, is that "a single hospital often times has a dozen or more systems within the network and those are constantly changing and vendors are constantly changing and typically are proprietary software."
There is no guarantee that disparate data systems (EMR, lab results, prescriptions, ICU, to give examples of data that may be collected on different systems within a single hospital) will function together, even within the network of a single hospital. "That's something that isn't even ever discussed when people are talking about just creating a big data system for a hospital."
Software companies make themselves more competitive in the current market by hoarding patient data and amassing proprietary data sets. While this makes sense from a business standpoint, this practice freezes data within companies that refuse to integrate with others. There is no economic incentive for companies to voluntarily open their proprietary data, but "without that," said Landman, "I don't know how much big data can truly accomplish."
Steve Hau, founder and CTO of Shareable Ink, echoed the idea that developing open data exchange was the key to unlocking big data's potential in an August 20 interview, saying that "data liquidity," or the ability of data to move from setting to setting without obstruction, "is the first step to get to data aggregation and then a meaningful view into population health."
Shareable Ink uses the implicit structure of clinical forms to organize written documentation recorded via digital pen or iPad. These devices send the documentation to the company's data center where it is processed into detailed, structured patient data. That is, liquid data. "We're essentially building and selling the picks and shovels to get to this big data gold rush"
Shareable Ink focuses on collecting highly detailed, highly structured data, not huge data sets. This choice relates to the debate surrounding big data over whether smaller, more detailed data sets yield better results than larger masses of heterogeneous, less-detailed data.
Focus of the goal
For Kaj Pedersen, CTO of Pharmacy OneSource, who recently spoke with Healthcare IT News, "the debate to some degree, in my mind, is probably futile. It's almost like people are taking positions one side or the other and what they're not necessarily looking at in the question is what is the outcome you are trying to achieve, who are you trying to serve, and what's the data that you need to support that?"
Analyzing massive sets of population data may generate better general understanding of diseases, but this means doctors asking if their particular patients match the population. Landman voiced the fear that "One of the risks and one thing we may see develop over the coming years as there is a push towards big data is the idea of the loss of the actual individual patient...I don't know anyone that can fundamentally tell you which one is better...and its ultimately, probably, going to be a balance between the two."
Unfortunately, and perhaps frustratingly, big data poses both great risks and great benefits, the negotiation of which stand to occupy the healthcare industry for the foreseeable future. The open source network ideal seems unrealistic for now (not least because of the heightened concern over data security and privacy provoked by the NSA scandal). But this talk of balance suggests a more pragmatic approach to the issue of big data. A culture of managed expectations is a necessary step in beginning to explore the greatest benefits of big data.
"In the last years...there were pretty much just the Star Trek notions of big data, of 'look what it could do,'" said Landman, "and now, more and more, we're talking about specifics, EMR integrations, market players...and I think that's fundamentally a lot more interesting"