The average Canadian lives in a world filled with almost limitless digital connections. Activity trackers monitor our steps and heart rate. Smartphones track our habits and browsing history to bring us personalized ads from retailers. Even home appliances are going “smart” with many now transferring information online.
In our digital age, information is the new currency. Huge volumes of complex data—often called big data—are collected by companies and governments who mine it for a better understanding of the needs and wants of the people they serve. It has been a winning formula for tech giants like Amazon and Google to quickly gain insights into their customers.
Now health innovators are taking the lessons learned in commerce and applying them to healthcare. “There is a complete revolution in healthcare on the way, and I don’t think many would disagree with that,” said Lawrence Richer, an associate dean of clinical research at the University of Alberta’s Faculty of Medicine & Dentistry.
Richer is one of several researchers at the U of A dipping his toe into the deep waters of big data with the belief it will provide new solutions to difficult problems. His own work involves a project that uses data to look for signals that would help emergency departments predict cases of stroke in children. For Richer, the future isn’t far away. “I was able to accurately predict those who presented with a headache and weak arm who were most likely to have something like a stroke,” he said. “(Big data) won’t replace care providers or physicians, but it certainly can augment our ability to make better choices, provide better care and stop doing things that aren’t of value.”
What is big data?
Big data is information that is huge and complex. It is messy. It is constant. In reality, that information has always been there for the taking but it’s only been in recent years, through advances in technology, that it could effectively be gathered, analyzed and acted upon in close to real time. While corporations have made huge strides in their ability to quickly use that information, its application in health care is still in its infancy. “We are going there but we are not there yet. That is our reality,” said Richer. “We are just at the beginning stages. We are just starting to digitize in many respects our regular transactions with patients.”
But for those who are working in the field, they see change coming very quickly. “Big data has been on the horizon for quite a while, at least 15 years. But I think it’s gained more popularity in the last five years,” said Rhonda Rosychuk, a professor with the U of A’s Department of Pediatrics. Rosychuk is a statistician whose research is focused on using administrative health data sets to look at emergency department visits in Alberta. In her work, she analyzes hundreds of thousands of records but hopes to soon link them to documentation of patient hospitalizations and physician claims—which would grow the scope of her work to involve millions of records.
She’s hopeful the additional information will help her gain important insights into how to direct patients to the most appropriate health-care service, but admits there is no certainty in how useful big data will be. “It has untapped potential as a driver of change in health but I think that potential has not been realized yet,” said Rosychuk. “There are a lot of issues—the quality of the data, the volume of the data. What does it mean? What kind of questions can you really answer with it?”
The truth is, no one quite knows what will come of the use of big data in healthcare. There is a lot of optimism but little certainty. The promise is tantalizing, though, and the landscape is changing week by week. “Will it open up stuff that we haven’t realized before? I don’t know. Ask me in a year’s time. Or even ask me in three months’ time,” said Padma Kaul, a professor of medicine with the U of A’s Division of Cardiology.
Kaul is an epidemiologist who does population-based health research. She is in the early stages of a project using big data to examine whether there are any predictive signals that would let physicians know if the mother or child are at risk of adverse perinatal or neonatal outcomes. “We are collecting so much detail on patients, we’re getting to the point where maybe we can look to the machines or to computers to recognize patterns that maybe we wouldn’t know of.”
According to Kaul, Alberta is in a unique position nationally and even globally in how researchers can benefit from big data. The province is distinct in Canada in how its healthcare system is organized with the single payer (public health care) and a single care provider (Alberta Health Services). The arrangement allows researchers to link data—records of hospitalizations, outpatient clinic visits, physicians visits, drug prescriptions—in a way that no other province can.
The unique arrangement is giving Kaul unprecedented amounts of information to sort through. Her team will be using computers enabled with machine learning to find patterns in the terabytes of available information. “My study is looking at records from 2005 to 2015. During that time, 300,000 women gave birth to about 500,000 babies,” said Kaul. “By the time you add up all the lab tests and their pharmaceutical data, we’re looking at about 70 million records. And then another 70 million records from when they saw their physicians. So you put all that data together and that’s like—boom! That’s big data!”
Analyzing it through traditional methods would pose a major challenge. According to Kaul, her team would have to simplify their efforts and possibly miss important information as a result of making the data more manageable for analysis. But with the aid of machine learning, individual pieces of data can be scrutinized and classified at a level of detail not possible by humans.
Very soon, health researchers in Alberta will have access to more information than ever before. Over the next few years the province is moving to implement a clinical information system called Connect Care—an electronic tool that will provide one central access point to patient information. That will be unprecedented, really, in most of the world to have that kind of data available at the population level,” said Richer. He warns, though, the collection of data comes with risks. “If EquiFax can be hacked, so can the health-care system.”
Big data, big risk
The amount of information being collected on every single person is growing by leaps and bounds. But as it does, the sensitivity of that information increases with it. “People may think that as an individual, you are not actually contributing all that much to this bigger data pile. But you may not actually have to contribute very much for somebody with access to all of this data to know quite a bit about you,” said Rosychuk. “I don’t think people are aware of how being a small part of data in a bigger picture can actually [allow someone with access to] make inferences about you or draw conclusions about you that you maybe don’t want.”
As an example, she points to the practice of just visiting a particular website associated with a disease. Rosychuk said there can be a probabilistic argument made that if you go to the site, you likely have the disease. “What if you had a sexually transmitted disease? Anything that maybe you want to have more privacy about, if there are a lot of other people who aren’t so private about it who end up visiting particular sites, inferences can be drawn about you.”
Even more concerning is the possibility of patient information being misappropriated and misused on a grand scale. “I think any false steps along the way—a hack, a release of data that shouldn’t have been released, a rogue researcher or clinician doing something they shouldn’t have done with the data—just decreases public confidence that we have their best interest at heart,” added Richer.“If I break into a doctor’s office, I can get a thousand charts. If I break into a data warehouse now, I get millions. So that’s the scale difference. That’s our biggest threat—doing this poorly. We could shut the whole thing down. One big mess-up and the public could say stop everything. And they would be right in saying so.”
With risk comes reward
While the potential risks are great, the rewards of big data could be even greater. Scientists believe the applications are almost limitless. “If you were to mix health data with education data, how could we better understand the health outcomes of children? If you were to mix health data with justice data, how could we better understand treatments in areas like mental health, for example?” said Richer. “We are at the cusp of that in Alberta and really working hard with engaged stakeholders who hold the data, wanting to see this happen. That’s the horizon we have.”
To get there, though, more than just technological advances need to happen. According to Richer, the major barrier that needs to be scaled isn’t computation power or the availability of data. It’s the availability of people who know how to work with it. “Our capacity to do that needs to go up. We need to triple the number of people who know how to do this and can do it well.”
Rosychuk agrees, but believes progression needs to happen on many fronts. “I believe that we need both. We need some more automated systems and then we need people who are smart and clever who can figure out alternative ways to identify patterns because they are looking at things in maybe a different way than a computer would.”
The ultimate goal is to have the lessons learned from information analytics complement the decisions made in the clinic. There are many avenues in which scientists believe that can happen, and when it does, there is a sense that the sky’s the limit for how much could be accomplished. “I think it’s definitely more a means of decision support,” said Thomas Covello, a resident at the U of A exploring the potential impact of big data on health care. “I think it’s going to lead to health-care processes becoming more protocol-driven and systemized rather than ad hoc. I think it’s going to be a good thing because it’s going to encourage evidence-based practice.”
“I think the promise with big data is that if you have evidence, you can solve any problem,” added Rosychuk. “This would be a way to find evidence cheaper, more easily and on a more representative group of people, as opposed to doing a small cohort or case control study that takes a lot of resources.”
While the information involved in big data is massive in scope, Kaul sees the impact being much more personal—affecting individuals by taking into account each one’s unique circumstances. “Precision medicine is all about big data in a sense,” said Kaul. “It is finding those associations between certain genotypes and disease and finding therapies specific to those genotypes. So I think you will see those kinds of strides as a result of big data.”
As for Richer, he sees a future—not far distant—in which big data and computer learning saves lives by alerting health-care providers at the point of care to issues in real time, often before they happen.
In some ways it’s already happening. He points to a system that has been tested in neonatal intensive care units in which computers monitor different signals to predict which infants have sepsis—before the bedside provider realizes there is a concern. “It allows the team to assess the baby and make care decisions based on those subtle changes that maybe weren’t apparent just by watching the screen,” said Richer. “It’s this predictive ability that I think is the real holy grail of using big data. That, to me, is the most valid application. It starts to compare now with Amazon being able to predict what you’re going to buy next and being quite good at it.”
There are big dreams for big data and the limits are still unknown. As the world continues to digitize and consolidate information, its applications are only going to grow. For health-care researchers at the U of A, a new day is definitely coming. Bring on the revolution.