How do we deliver the right healthcare data to the right person at the right time?
In any discussion of medical data, you are likely to hear two concepts more than any other: the silo and the “blue button.” A silo is any structure that prioritizes ownership over use. The blue button is a metaphorical object that allows anyone anywhere who needs healthcare information to get it, regardless of “ownership.”
Unfortunately for all, the former continues to dominate today.
Any examination of the ever-increasing volume of healthcare data has to include the question, how can we move from an incomplete and isolated system of medical data to one in which all the parts are united, and users can summon up that information they need without compromising patient privacy?
In other words, how do we shift from silos to the blue button?
Here, experts recommend real hands-on, ground-up actions that can help everyone involved get what they need, starting with the systems we currently have in place.
A landscape of silos
Every medical examination, study, prescription, and policy change creates data. Add millions of new apps, wearable devices, and connected medical instruments each year, and it’s clear that healthcare is newly rich in data—by one estimate, the field is generating as much as 30% of the world’s data volume.
More data and better analysis should lead straight to better outcomes—in theory. In practice, though, there’s a big question standing in the way, according to Saurabh Gombar, Co-Founder and Chief Medical Officer at data-based physician consultation company Atropos Health. And that question is “Where is the data located?”
“Right now, most of the data is siloed,” said Gombar. “Hospital systems tend to have their own electronic health records systems.” A substantial amount of the data found in these systems is structured, and therefore easy to search and analyze. But a lot of it remains unstructured, such as physician notes.
“That’s very rich information, but it’s stored as plain text; it’s not structured in any specific way. That’s the health system’s view of an individual. And often, it’s siloed to that individual provider. Most systems do not talk to each other, so it’s hard to follow a new patient from a previous system.”
Individual health information like what we get from siloed sources, such as exercise devices, is even less likely to be included in aggregate of data to draw from.
The blue button and the deadbolt
The original blue button was begun by a handful of agencies in the U.S. government with the goal of making health information available as needed across those departments.
To extend this experiment radically far afield, we need to define how a person becomes the right person in terms of data and time. Gombar believes that transition happens at the juncture of decision-making, and that context in turn helps us preserve one aspect of data without which there can be no “right” anything: privacy.
“We want to make the correct data available, for the correct user,” said Gombar. “So the private health information [can be used by] the physician, or the nurse, or the practitioner [because] they are the right and appropriate user of the information in the context of making decisions.“
“So as long as we’re keeping the right information, for the right user, I think we automatically meet the privacy requirements,” he said. “The danger comes when we start opening up all of the health information, for choices that have nothing to do with the specific patient.”
Additional obligations for privacy include legal and regulatory requirements. In the United States these include the demanding strictures of the Health Insurance Portability and Accountability Act (HIPAA); in Europe the GDPR plays a similar role. Another means of enabling companies to maintain privacy is the use of AI, which can help guard against intrusion. Data sets used to train an AI system generally do not require identifying information. In the end, finding personal information is always possible, even with anonymization techniques applied. It is the responsibility of those managing that information to make it a little harder than hackers are willing to work.
However, regulation and data anonymization do help to increase patient trust, and that is integral to the sharing of data.
From transactional to analytical
Todd Crosslin, Snowflake’s Industry Principal, Healthcare and Life Sciences, described a way to make more data more useful—how we might press that blue button—as a transition of data from transactional to analytical. As data makes that transition it becomes information, and information allows us to make smarter health decisions.
“There are so many medical devices that are just in phase one of data, employing basic analytics at best,” said Crosslin. You wear a fitness monitor, and it tracks your activity—you walked 1,000 steps or climbed the equivalent of 35 staircases— but the device company doesn’t go any further and analyze the data to give suggestions. You get some basic intelligence, but it’s raw and transactional.
“But what we want to do is push them to phase two: advanced analytics and data science,” he said.
When a user transforms data into information and that is used to understand medical issues, you have created research. This is what Crosslin calls transition from real-world data to real-world evidence.
“There’s a key differentiator there,” Crosslin noted. “Real-world data is data. Real-world evidence is analytics on that data.” That transition from data to evidence is a big step toward a global blue button.
A personal example
Crosslin provided a personal example of how powerful that blue button could be. He went to see his doctor because his heart had begun to race. His doctor asked him a host of questions and administered another host of tests. Nothing. Finally, his doctor noticed acid reflux on his chart and thought to ask him if he ever took omeprazole. Rather surprised, Crosslin said he had; he had taken Prilosec, whose active ingredient is omeprazole, for his acid reflux. The doctor had read about that particular ingredient depleting magnesium in men, causing the symptoms he experienced.
If Crosslin’s doctor hadn’t thought to ask that question, or hadn’t come across that poorly documented side effect, Crosslin would have in all likelihood continued to suffer. On the other hand, if his doctor could have pushed a blue button connected to all the pertinent data, that included over-the-counter pharmacy information, he could have diagnosed his condition much more quickly.
Enabling a global blue button is not an abstract question of process. It is not solely a diagnostic issue, a public policy matter, nor a financial concern. Depending on the patient, it could be a matter of life and death. The responsibility for pushing its creation forward belongs to all of us, but those deeply involved in understanding and using data may be said to shoulder a more significant burden in its development and implementation. It’s time we talk about it—immediately, loudly, and publicly.