- When it comes to healthcare big data analytics, finding the pot of gold at the end of the rainbow isn’t exactly an easy task. Even the process of identifying if there is a pot of gold – and how to plot a course towards the rainbow before planning out the best way to follow it – can be monumentally difficult for organizations mired down in EHR optimization projects, meaningful use, payment reforms, and budding population health management programs.
But there are definitely rewards to be had for healthcare organizations that keep their eyes on the skies, says David Delaney, MD, Chief Medical Officer and leader of SAP’s US healthcare division, and the rainbow looks very similar to a coiled strand of human DNA.
Healthcare big data analytics should be leading providers towards the operationalization of precision medicine, says Delaney. The journey will not only require a significant investment in new analytics and data management technologies, but also a perspective shift for providers who are still struggling with the best ways to extract value from an overwhelming flood of patient information.
While “precision medicine” and “personalized medicine” are often used interchangeably to describe the ways providers are focusing on the individual patient and his or her responses to particular therapies, there is actually a notable difference between what these two buzz words really mean.
“I characterize ‘precision medicine’ as how we’re changing the diagnosis process and clinical decision-making to be based on genomic factors,” Delaney said in an interview with HealthITAnalytics.com.
“Even more broadly, all the ‘-omic’ factors that genetically control us, from the controlled DNA sequences to what actually gets transcribed into proteins to the output of those proteins in the metabolites. That, in combination with social, lifestyle, cultural factors, gives us precision, and informs all those nuanced decision for the patient.”
Meanwhile, “the phrase ‘personalized medicine’ tends to conjure up the idea of synthesizing a particular molecule that's a tailored treatment for a particular person,” he continued. “Precision medicine uses the same types of information, but in ways that guide therapy and therapeutics to create much more fine-grained groupings of patients that goes beyond the population health management capabilities we have today.”
The distinction is an important one as the healthcare industry delves deeper into the process of tailoring treatments for patients based on their specific attributes, lifestyle factors, and genetic makeup. As providers begin to look at population health management from both a macro and micro perspective, being able to categorize – and appropriately reimburse – individualized treatments will become increasingly important.
“Right now, all we typically offer is one-size-fits all treatments,” said Delaney. “If you have this illness, you get this treatment. Sometimes we have data from randomized controlled trials that help us be a little more precise, but it’s still like buying a t-shirt. You've got your small, medium, large, and extra-large, which is better than one-size-fits-all, but not by much.”
That approach is going out of fashion very quickly, however, as healthcare’s big data analytics competencies improve. More and more organizations are recognizing the importance of collecting and synthesizing information outside of the purely clinical realm.
The Internet of Things is adding patient-generated health data from wearables and medical devices to the mix; lifestyle and socioeconomic data from social media, public health organizations, and behavioral health providers are steadily becoming valuable additions to the EHR. Open notes allow patients to contribute to their own story, and innovative new cognitive computing technologies are helping provider organizations make sense of it all.
By correlating this “softer” patient-related data with clinical notes, genome sequencing, and the outcomes of previous clinical trials, some medical specialties are undergoing explosively innovative revolutions. Oncology has seen especially notable early success, and has become the focal point for many precision medicine initiatives as providers start to change the very definition of cancer.
“Cancers used to be treated based on the organ in which they occurred,” Delaney explained. “Lung cancer was lung cancer. There was one type and you treated with a handful of agents that were good against lung cancer. Same thing with the pancreas or kidneys or anything else. And some patients did really well and got cured, some people did okay and lived a little longer, some people did very badly and died quickly.”
“Flash forward a few years, and now we look at cancer on a molecular level, based on the type of mutation and the signatures of the tumors,” he continued. “So, it turns out, lung cancer isn't lung cancer, but it's a whole bunch of various, individual, discrete illnesses based on where the mutation occurred and what molecular markers there are.”
“So we’re moving away from treating cancer based on the organ where the malignancy. Now, you might have a lung cancer that's treated similarly to a pancreatic cancer because the origin of the mutation is more similar to pancreatic cancers, despite being in a different organ.”
This shift in perspective is one of the immediate benefits of taking a precision medicine approach to something like cancer care, he said, and recognizing the sheer depth of what clinicians don’t yet know about deadly diseases is just the beginning.
“It's not an on or off thing – it's very much a Zen-like journey of becoming,” said Delaney. “We’re pushing back the walls of science, beginning to understand in a finer-grained fashion what the underlying causes of illness are, and then understanding the optimal treatments for them.”
While precision medicine seems like the exclusive domain of the big academic medical centers and research powerhouses, due to the expense of securing research staff, hosting large-scale clinical trials, and developing the technologies required to support these activities, Delaney doesn’t believe that smaller organizations will be left out of the loop forever.
“The technology underlying precision medicine is getting simpler, more mature, and more affordable,” he said. “It’s not so affordable that every small hospital can have the same kind of computing capabilities as a major academic medical center. But the gap is narrowing.”
It might not be narrow enough for the majority of smaller providers, however, who simply do not process the volume of patients required to build a meaningful data set. “You have to have a large enough dataset to support what you're trying to study,” Delaney said.
“If you have something pretty common, like heart disease or diabetes, you’ll have enough patients, even in a small hospital, to take a meaningful look at the data. But if you have a more unusual illness and you only have ten patients in the hospital being treated for that condition, you’re just not going to have enough information to do the research.”
“So smaller organizations are always going to be a disadvantage, somewhat, in that regard. They’re just not going to have the ability to look at rare diseases the way a large medical center or integrated delivery network can. And even then, there are some conditions that are so rare that you really need a national or international registry to be able to consolidate enough patients to get meaningful data about it.”
The White House is hoping to fill that need by creating a national database with genomic and other information on at least one million patients. But the process of designing such a large system is a difficult one, and the NIH is still waiting to get approval for funding that will get the project off the ground.
In the meantime, healthcare providers and research centers who are pushing the boundaries of precision medicine on their own dime need to take a harder look at just how well they are representing the patients they hope to treat.
Only a very small portion of oncology patients, for example, are ever enrolled in controlled clinical trials that produce the literature used for decision-making across the healthcare system, said Dr. Clifford A. Hudis, MD, FACP, a medical oncologist at Memorial Sloan Kettering Cancer Center and past president of the American Society of Clinical Oncology (ASCO) in an interview earlier this year.
SAP and ASCO have been collaborating on the development of the CancerLinQ platform, which is designed to leverage previously unavailable patient data to craft a more detailed and accurate portrait of what might work best for specific subpopulations.
“Only about 3 percent of people with cancer actually participate in research,” Hudis said. “Ninety-seven percent of the experiences we get while treating cancer are outside the clinical research setting, and up until now we couldn’t mine any of it. With rare exception, nothing is truly learned from it.”
“If I ask my patients what they think happens with their experience, they all think they’re contributing to our learning. Even patients having nothing to do with a clinical trial think their doctors are learning from the experience of treating them. Sadly, that’s never been particularly true.”
And the patients that do get to contribute to research are often not representative of the majority of cancer cases, Delaney added, which limits the effectiveness and applicability of the results.
“When you look at the three or five percent of patients who are included in clinical trials, they’re all, by definition, very vanilla patients,” he said. “You’re trying to control the variables in the study and get two populations who are very similar to each other, except for the treatment they’re given.”
“But in the real world, what happens when an oncology patient comes to see you?” Delaney asked. “It might be a 78-year-old person who has lung cancer, but he has a history of prostate cancer in the past. He has CHF and diabetes. That's a very common patient. But when you’re looking at clinical trials to find the best treatment for him, you’re not really looking at what happened to patients like him. So everything you’re doing is essentially going to be off-label.”
“But the data on those patients is out there. The other 95 percent of patients have been treated, and the data exists in the EHR and registries. So, why is it we can't pull it together and actually make a meaningful recommendation to the oncologist?”
To some degree, new advances in healthcare big data analytics technologies are allowing providers to do just that. From IBM’s Watson cognitive computing platform to semantic data lakes filling up at organizations like Montefiore Medical Center, providers are slowly gaining the ability to get answers to very specific clinical questions that may not have ever been asked before.
This incredible personalization requires a new approach to the data itself, Delaney states. Traditional relational databases that demand a high degree of pre-planning to be effective will not provide the power, flexibility, and ingenuity that precision medicine needs.
“When you look at the traditional way of doing big data analytics, it's very slow and plodding,” he said. “We have relational databases, and all these data warehouses which require you to have an understanding of the data when you’re building them. So when you’re trying to discover new relationships, you end up making a lot of false starts. When you model it out, nothing is quite right. You do data mining and research and then you refactor, remodel the data, and you move it around.”
“When you're working with large datasets, these are very large, multi-month cycles. That’s very, very good when you want to create reports based on information you already know you have, but it’s not so good when you're trying to understand novel relationships and create insight and value.”
“And so, that's where things like in-memory technology enable you to have a very simplified model of the data,” he continued. “You can essentially apply lenses to view it in a number of different ways, and then you can understand what is valuable and how to extract those insights. That’s one key thing you need for precision medicine. You need to be able to fall back to a very liquid data format, so that you can mine it very easily and discover new things.”
These technologies may be quite a bit farther along the rainbow than most providers can reach at the moment, and a deep dive into precision medicine might not be on the agenda for community hospitals and other providers still trying to come to terms with productivity losses, workflow frustrations, and patient safety issues caused by bringing computers into the consult room.
“Right now, precision medicine is an ‘ought to do’ initiative,” Delaney acknowledges. “It doesn’t have the same urgency as getting through your day and getting paid and making your mortgage payment. We need to move it up the hierarchy, because it’s good for patients. And it’s got to be part of the shift from volume to value, which is a very complex movement.”
Precision medicine is an integral part of the accountable care environment, Delaney argues. “Value-based care can't really occur in a meaningful context until we begin to apply precision medicine much more broadly,” he said. “Because the only way we're going to be able to pay for value is if we understand how to deliver it. Precision medicine is an opportunity to do that.”
“It’s going to be a slow and painful process to get there, but it will accelerate over time as a few organizations start delivering value and start figuring out how to do this well,” he predicts. “They'll be able to look at these very precisely defined populations of patients and say, ‘Which subpopulation is doing well on this kind of treatment? And how can we begin to make this reproducible?’"
“I think once patients begin to understand that there is a better way out there, that there's a more data-driven way out there, they will begin to push for it, too. So will providers. They have been forced to make decisions based on limited datasets for too long, and they’ve become immune to it. But when they realize there’s a better way, I think expectations will shift.”
The shift is already apparent in many areas of the country, where value-based reimbursement is becoming the norm and patient satisfaction scores, formal and informal, are starting to have a major impact on business practices and organizational rankings.
Patients are starting to expect that the streams of data they are sending through their iPhones, Apple Watches and FitBits are being viewed and used for decision-making that will be unique to the situation at hand, which places a serious onus on providers to treat this data as an integral factor in the care process.
This can be tough on clinicians who are not yet prepared to make the adjustment, or who don’t have the EHR and population health management technology in place to make sense of this barrage of new data points. But as providers start to develop new strategies to turn data into meaningful information, they may start to recognize the value of the insights hidden within the reams of sleep statistics, calorie counts, and blood sugar fluctuations.
Precision medicine has to start somewhere, and learning how to accept and manage patient-generated health data may help health organizations build the bridge to more sophisticated genomics, better clinical trials, and increasingly effective, personalized treatments for cancer and other serious diseases. The process might not be an easy one, but there is a great deal of gold in the precision medicine pot if healthcare organizations commit to the voyage.
“I love the fact that we're setting out on the journey,” said Delaney. “We might not be getting everything perfect right at the beginning, but we’re starting to make progress, and that’s the important thing. We are just beginning to accrue the data and to shift to a new way of thinking. All this data has value, and if people begin to contribute it, together we can start to do great things.”