EMRs provide a rich source of data points on human health, but interconnectivity remains a hurdle to harnessing the power.
All that information dermatologists are plugging into their electronic medical records (EMRs) might seem tedious, unnecessary and irrelevant, but there’s an important flip side that could mean better medicine and reimbursement down the road.
“A dermatologist could ask, why should I record what specific medications my patients are currently taking if I am only treating their basal cell cancer?” said Kavita Yang Sarin, M.D., Ph.D., clinical assistant professor of dermatology, Stanford University Medical Center, Stanford, Calif. “However, a researcher can use this data over millions of patients to find a new association with a specific medication and increased risk of skin cancer. In this way, the individual patient and physician contributes more broadly to improving human health.”
Big data can be used to fine-tune interventions to manage increasing numbers of patients at a lower cost, according to Mark Seraly, M.D., dermatologist in McMurray, Penn. In addition to in-office patients, Dr. Seraly has more than 1,200 online patients each year, through a business he founded, DermatologistOnCall, which is used by more than 150 U.S. dermatologists.
“In an office setting, [an] EMR platform, for example, is used to track melanoma patients to ensure compliance with follow-up visits, coordination of care with the patient’s other healthcare providers and lab testing,” Dr. Seraly says. “In teledermatology … studying big data can help us better understand consumer behaviors as [they relate] to virtual health - how data can help us deliver a ‘personal’ connection in an often impersonal online setting. It can also help us optimize workflows and processes, such as predicting what percent of online patients will need to be referred to an in-office consultation; predicting what percent of online patients will need prescription care; and measuring time-to-complete for online cases.”
Big data figures prominently in healthcare’s new value-based payment system, called the Merit-based Incentive Payment System, which goes into effect in 2019, said Michael Sherling, M.D., dermatologist, cofounder and chief medical officer of Boca Raton, Fla.-based Modernizing Medicine, and co-developer of EMA, a cloud-based, iPad native EMR system, which powers 35% of the U.S. dermatology market.
“What we do in 2017 depends on that 2019 reimbursement,” Dr. Sherling says.
With the Merit-based Incentive Payment System, dermatologists will get paid based on the data they put in and how it measures their quality of care; cost compared to other physicians; as well as something called practice improvement, which is the ability for physicians to use their own data to manage population health, Dr. Sherling says.
NEXT: Data-driven payments
“So, the way that physicians get paid depends on the [EMR] being able to capture the data and the physicians being able to see their own data over time,” he says.
In essence, EMA collects the structured data that physicians use to document a visit. That data then gets put into a business intelligence tool for the physician, so the physician can see their own data at a much higher level. Modernizing Medicine launched such a technology - its ModMed Analytics tool - at this year’s American Academy of Dermatology annual meeting.
“Instead of just looking at it for every patient, they can look at their data for an entire year for all their patients. And then they can see benchmarks for how they perform, compared to the entire user-base of EMA, as well as accepted Medicare benchmarks. They can see if they’re ahead of the curve or if there are things they need to work on in terms of managing population health,” Dr. Sherling says.
Big data in medicine involves the large scale analysis of data to improve human health. EMRs provide a rich source containing millions of data points on human health and disease on millions of patients, Dr. Sarin says.
The positive end result of all that data is not yet reality in healthcare, however, according to Dexter Hadley, M.D., Ph.D., assistant professor, Institute for Computational Health Sciences, at University of California, San Francisco. In fact, healthcare is considerably behind other areas, such as banking and retail, when it comes to harnessing big data’s power.
“For instance, you can go on Google and type in dog or cat, and you will find tons of dogs and cats, with labels for each one. The computer is able to classify each picture as a dog or cat,” Dr. Hadley says.
Computer algorithms have outperformed humans since 2012 in their ability to classify dogs, according to Dr. Hadley.
How does Google do that? Essentially, by training the algorithms with massive amounts of data.
“They’ve gotten a lot of their data from Facebook, for example, where people label their dogs. That’s not going to work in healthcare. In other words, people don’t label their nevi on Facebook,” Dr. Hadley says.
NEXT: Derm-specific disease applications
Dr. Hadley recently received UCSF funding to develop an algorithm that can predict skin cancer from images taken with a smart phone. The accuracy, he says, has to be at least as high, if not higher, as predictions by dermatologists.
Going from little data (at the practice level) to big data ready for use and mining is tedious, Dr. Hadley says. That’s because even at an institution as large as UCSF, mole images that doctors capture each month are stored on a server, which isn’t part of any other data base. In other words, the data exists, but it’s just sitting there. Dr. Hadley and his colleagues are going through it manually, in their attempt to amass thousands of images labeled with pathology this year. With those images, which include cancerous and benign moles, Dr. Hadley can essentially train an algorithm to triage skin cancer risk. The resulting technology could help dermatologists determine if the moles they see are atypical or pre-melanoma.
To move this big data into the digital health realm will require that not only dermatologists have an integrated way to send images, but also patients. Patients can drive big data by taking selfies of their moles, which will document a mole’s evolution.
“That way, we engage the patient, we collect the organized data we need and can do high-level research on it. We can give an alert to the doctor that a mole becomes a concern, based on the [thousands of] images on our database,” Dr. Hadley says.
The very nature of an EMR is to capture information; not push it out for the better good, according to Christopher Wynder, Ph.D., client services director for ThinkDox, an information management company.
“When we talk about data or information, the context or meaning comes from the metadata. The best examples of metadata in healthcare are ICD-10 and HL7, a pair of coding systems,” Dr. Wynder says.
ThinkDox supports Laserfiche enterprise content management (ECM) for electronic document/content/records software implementations. Laserfiche, according to Dr. Wynder, has the flexibility to add metadata from any source- including the ICD-10 and HL7 coding systems to any piece of content (images, test results, emails) and match it to the patient data in the EMR.
“Products that have strong records management certifications, like a Laserfiche, are starting to be used to replace legacy EMR systems in the UK,” Dr. Wynder says.
One of the hurdles to harnessing big data for use is that while the government has mandated EMR use, it didn’t define interoperability.
“That’s a huge problem because there is no standardized way to mine this data. Right now, EMR companies are making the technology proprietary, and that makes it even more difficult to be interoperable across institutions generating data sets that are well-described and labeled,” Dr. Hadley says.
The challenge, according to Dr. Hadley, is researchers can’t survey a population closely enough to pick up the evolution of melanoma from skin moles to melanoma.
“We can’t track this because we don’t have images across a spectrum of premalignant to malignant for patients. Neither do we have molecules - either RNA or DNA,” he says. “That’s very hard to do unless you’ve had some kind of personalized surveillance.”
Images from personalized patient surveillance could lead to alerts, where patients take a next step.
Glucometers are a great example of how this kind of personalized surveillance technology has worked in medicine, according to Dr. Hadley. In essence, a doctor prescribes a patient to follow their blood sugar at home with a glucometer, and the doctor acts on that information to better manage their patients' diabetes.
“I want to create a ‘glucometer’ for melanoma,” Dr. Hadley says.
NEXT: Overcoming hurdles in healthcare data
The good news is EMA and other leading EMR providers are working on overcoming interoperability issues concerning healthcare data.
“We’re … members of the CommonWell Health Alliance, which is the partnership for interoperability by the leading EMR companies to share data, and we’re interested in working with the Academy with their DataDerm initiative,” Dr. Sherling says.
Dermatologists can enroll in the American Academy of Dermatology's DataDerm https://www.aad.org/practice-tools/dataderm clinical data registry, which will allow them to contribute data, as well as meaningful use criteria, thus improving dermatologists’ pay, according to Dr. Sarin.
At the practice level, dermatologists can help pave the way for useable big data by using standardized templates on their EMRs, which allow researchers to more easily parse for data, according to Dr. Sarin.
“By keeping standardized, careful EMR records, each patient and doctor contributes an individual piece of data that can be used to discover scientific and clinical advances to improve human health,” Dr. Sarin says.
By the same token, recording incorrect information can be detrimental to big data discovery. So it’s important that dermatologists accurately and precisely record clinical data for robust big data analysis, according to Dr. Sarin.
Dr. Sarin reports no relevant disclosures. Dr. Hadley is part of the company, SkinIQ, which is founded through the QB3 incubator at UCSF, and is working to put this digital health tech into the hands of doctors and their patients. Drs. Sherling and Wynder own companies in this space.