adimas - Fotolia
The potential benefits of genomic data analysis in healthcare are undeniable. The list includes reducing repeat testing, predicting what illnesses someone may experience in the future, providing more personalized care to patients and more. But the list of challenges is, unfortunately, just as long.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Experts agree that solutions to these challenges need to be sorted out before the full potential of genomic data analysis can be realized in healthcare.
The problem with genomic data
Keith Stewart, director at the Center for Individualized Medicine at the Mayo Clinic in Scottsdale, Ariz., said at Health Datapalooza in Washington D.C. in April that some of these barriers include patient wariness, a lack of education in this field among physicians, a shortage of genetic counseling and healthcare payers not yet wanting to pay for genetic tests.
But improvements to the technology are also needed, Stewart said, especially when it comes to dealing with data. We're talking about handling something like four petabytes of data here.
Not only that, another issue is figuring out how to present genomic data in a meaningful way to the physician.
Keith Stewartdirector at the Center for Individualized Medicine at the Mayo Clinic
"How do you put a whole genome into the electronic medical record? Where do you store information?" Stewart said. "How do you go back and reanalyze that every year as science starts to tell you new information about the genome, so we don't sequence some of these genomes and let it fester in a cupboard? [It needs to be] an active, live resource through their life, not a static test that's done once."
Stewart explained that, presently, genomic test results are printed out on a single piece of paper. That piece of paper is then scanned into the EHR as a PDF file.
"The physician is probably never going to be able to find it a year after it's scanned into the report, and there is no way for most electronic health records to support genomic results today," he said.
Stewart added that the lack of clinical decision support when it comes to genomics is also an issue. Having data presented to a provider based off of a patient's genomic data that basically says, "'mutation X equals activity Y' is not available yet," Stewart said.
Integrating personalized data
To truly benefit from genomic data analysis, it's not just about genomic data, said Emma Huang, associate director of research and development at Janssen Pharmaceuticals in San Francisco, at Health Datapalooza. It's about linking that genomic data up with rich phenotypic data, clinical data and the EHR.
This is particularly crucial, Huang said, if the goal is to achieve predictive genomics. And in order to achieve that, data on peoples' lifestyles will also need to be collected, whether from sensors or wearables.
Huang added that information about the environment of the patient is also important, "so we know what's happening around people, so that we can predict the probability of disease."
Delivering data to the patient
Huang said that another challenge genomic data analysis faces is making the data not only freely available to patients, but also understandable. This is where analytics is key, Huang said.
"[Analytics] means really moving from not just generating the data, [to getting] insight from the data," she said.
To Huang, this is what data liberation means.
"It's not necessarily making data freely available to everyone. It's really making the insight from the data much easier [to understand] and making that available," Huang said.
The key to integration: Data movement
Mark Dunnenberger, director of pharmacogenomics at the Center for Molecular Medicine at NorthShore University Health System in Chicago, agrees with Huang that the main challenge is data integration.
"Genomics needs to be integrated with all of the other clinical data that's available for the patient," he said at Health Datapalooza. "One of the ways we have to do this is [through] many steps of translation, and this is where we get data movement."
Dunnenberger explained that, at present, the various data about someone's genome are processed in different places. For example, one laboratory may provide a patient's genotype, but a different facility would have to provide that same patient's phenotype. All that information should reside in one place so that physicians don't have to go looking for it. And in order for that to happen, standards are required.
Furthermore, physicians shouldn't have to go looking for this information at all. It should be presented to them point of care, when they need it. Dunnenberger said this will most likely appear in the form of some kind of alert.
"We're [made up of] tens of thousands of genes," he said. "We're struggling with, how do you build this logic when we're taking into account four, five and six genes at a time when you have to have performance capabilities to deliver this in microseconds? Well, it might be an interruptive alert."
This alert would have information on that patient's genetics, such as what medications do not interact well with the patient and possible diseases they may be at risk of developing.
In Dunnenberger's experience, he's found that some clinicians simply want a short and straight-to-the-point alert, and others want a longer explanation of the results. He has found that giving physicians a choice of the length and detail of the alert on genomic information is helpful.
Making genomics possible in healthcare using visual data analytics
Precision medicine may not always be so precise when treating cancer
Genomics may encourage use of the personal health record