Get fresh updates from Hortonworks by email

Once a month, receive latest insights, trends, analytics, offering information and knowledge of the Big Data.

cta

Get Started

cloud

Ready to Get Started?

Download sandbox

How can we help you?

closeClose button
April 29, 2015
prev slideNext slide

HDP for Healthcare Providers: Common Data Challenges

In today’s healthcare industry, shifting reimbursement models and increasing costs for supplies and labor come at the same time as mandates to improve care delivery while lowering costs. Improved healthcare outcomes should usually come at a higher cost, but more and better data can drive insights and efficiencies that help with both of those opposing pressures—helping care providers create new ways to both practice medicine and do business.

In her blog post entitled “Top 5 Health Care Trends to Watch in 2015,” Susan DeVore described this year’s top healthcare challenges and how the industry is addressing them.

Here is how Ms. DeVore describes the five trends:

  1. The Year of Living Interoperably” (overcoming data fragmentation)

    …From electronic health records (EHRs) to clinical measures and decision support tools, providers are inundated with new technologies that automate processes and capture new types of data. However, these systems are limited in their potential because they don’t all “talk” to one another…

  2. What Goes Up Must Come Down” (the high cost of specialty drugs)

    … Specialty drugs are expensive to develop and target a very small subset of the population…Just 1-3 percent of the population uses specialty drugs, yet spending for these therapies will account for 50 percent of total U.S. drug expenditures by 2018…

  3. Innovation Nation” (state-driven innovation in healthcare models)

    … Although almost all attention has focused on federal programs designed to incent greater care coordination, disease management and preventive care, innovations are starting to take root in the states, too…

  4. “Let the Sun Shine” (greater price transparency)

    …More and more health care data will be made available to public scrutiny in 2015. The newest area of transparency is financial, with Medicare’s release of physician procedure charges across the nation, as well as new data comparing insurance plan premiums and benefits through the health insurance marketplace…

  5. “Bundled Payment Takeover” (shifting reimbursement models)

    …Along with all the attention on accountable care organizations (ACOs) as the leading model for care transformation, this year will see a rapid expansion in bundled payment. Bundled payments group all the services patients receive during an episode of care…under a single target fee; if providers are able to provide necessary services more efficiently and with a higher quality outcome, they keep a portion of the savings generated…

Some of the world’s leading hospitals and healthcare systems use Hortonworks Data Platform (HDP) and subscribe with Hortonworks to address the challenges posed by these five trends for 2015. We work with these innovative healthcare providers to build advanced analytic applications for a single view of their patients, predictive analytics for cost control and quality improvement along with more effective data discovery to gain visibility into new ways to improve care and reduce costs.

At Hortonworks, we are experts with these data driven applications because (in addition to healthcare) we’ve partnered with more than 300 HDP clients to solve similar analytical challenges in other industries such as: insurance, telecommunications, retail, oil and gas, advertising and manufacturing.

Challenges Building a Single View of the Patient for New Reimbursement Models

The dual mandates of reducing costs while improving outcomes puts enormous pressure on traditional fee-for-service reimbursement models, leading providers to require more granular visibility into blended models and bundled payments. Many of these approaches treat patient care in a more retail-like context.

Like the many Hortonworks customers in the retail industry, our healthcare subscribers strive to create a 360-degree view—“a single view”—of their patients, but they are frustrated to find that innovative approaches to reimbursement typically require more data to validate the new models or to certify success. Medical centers pilot these new models alongside their existing reimbursement processes, increasing the strain on existing data architectures, with the risk of further fragmenting a provider’s view of its patients.

These care providers can rely on HDP’s centralized architecture to store and process the quantity and variety of data they need to validate new reimbursement models with a single view of patients. With HDP they can combine data from different points in the chain of care. They can analyze longitudinal patient histories over multiple touch points and episodes of care across providers, clinics, pharmacies and specialty providers. Our healthcare customers can also store and process data from less structured sources that were not suited to storage in a traditional column-and-row database.

In an earlier guest blog post “Journey to a Healthcare Data Lake: Hadoop at Mercy” Paul Boal, Director of Data Management and Analytics at Mercy, described how his team partnered with Hortonworks to build a single view of its patients. Boal described how Hadoop helped Mercy overcome challenges of data structure and economics to build this single view through their Mercy Data Library and patient vitals project:

The strength of Hadoop as a data platform is its ability to ingest and combine data sets from all these sources and formats. The combination of all of these data sets in a common platform enables us to ask questions that we weren’t able to ask previously, and we can ask those on an increasingly larger scale. Because of the low cost of storage on the platform, we can store information that we might have otherwise ignored if it were at a higher storage cost.”

Challenges Creating Predictive Analytics to Control Supply Chain and Labor Costs

Changes in healthcare regulations and mandates around population health only increase the pressure on healthcare entities to control costs. Much of that cost resides in the medical supply chain and with labor costs, and a lack of detailed, reliable, real-time data is a major obstacle to reducing those costs. Apache Hadoop helps provide the data for predictive analytics for cost savings.

First, lets look at the supply chain. Hortonworks customer Cardinal Health provides an excellent example of how capturing big data can improve outcomes and reduce costs through predictive analytics. Cardinal Health optimizes its medical supply chain with predictive analytics powered by a combined solution with Teradata and Hortonworks Data Platform (HDP). Watch the video and hear Neeraj Kumar, Cardinal Health’s Vice President of Information Management and Analytics, describe how predictive analytics helps Cardinal Health anticipate the path of flu epidemics and deliver medicine to the most appropriate locations:

Long-time Hortonworks customer UC Irvine Health (UCIH) provides another example of predictive analytics. They ingest real-time streaming patient data, which they analyze in combination with historical data on patient outcomes to gain better visibility into post discharge patient behavior. The hospital uses that insight to reduce re-admittance rates for cardiac patients.

Due to new readmission mandates, UCIH’s goal is to predict the likelihood of hospital re-admittance within 30 days after discharge. Patients with congestive heart failure have a tendency to build up fluid, which causes them to gain weight. Rapid weight gain over a 1-2 day period alerts caregivers that something is wrong and that the patient should see their doctor.

UCIH collaborated with medical device integration partner, iSirona, to develop a program that sends those heart patients home with a scale and instructions to weigh themselves at scheduled times. The weight data is wirelessly transmitted to Hadoop where algorithms determine which weight changes indicate risk of re-admittance. The system notifies clinicians about only those cases. All home monitoring data will be viewable in the EMR via an API to Hadoop.

This is a perfect example of how predictive analytics and preventative care with HDP meets two elusive goals at once: better patient health with lower costs for patients and providers.

Challenges with Discovering Data to Improve Care While Reducing Costs

Hadoop-based applications for data discovery also help meet those dual requirements of cost reduction and improved health outcomes. With greater data visibility, our healthcare customers can reduce waste and discover new clinical approaches.

One of the main challenges with existing EMR systems is that they were architected for high-fidelity data for patient-physician interactions at the level of each individual patient. Their intended purpose was not to explore and summarize data across patients. Most data storage and access platforms specialize on either individual records or data aggregation, but not on both at once.

Hortonworks Data Platform’s YARN-based architecture lets medical professionals analyze big data in many different ways at once—giving them both deep, detailed insight into the particulars of one patient and also the ability to identify patterns across millions of different patients (while still preserving the underlying detail on each individual).

For example, one Hortonworks subscriber is a part of the Canadian, publicly funded healthcare system. Data scientists at this research hospital have access to complete, lifelong healthcare data on several million individuals. Even so, those robust datasets are fragmented across many databases, and so it is difficult to discover patterns that span multiple datasets on different platforms. It was difficult to combine the data at sufficient scale to improve the delivery of healthcare.

The Director of Research at this hospital led the analytics team to adopt Hadoop for more efficient data discovery around different research hypotheses. He described the power of the big data approach this way:

For example, we’re trying to define a population of people with epilepsy from this data set and one of the administrative case definitions for epilepsy is the person needs to have a medical event that the doctor coded as an intervention to deal with epilepsy AND the presence of an anti-epileptic drug within a six-month window of that event. Well, if you’re talking about a table with a billion records in it and you try to do date-diff calculations on it, it doesn’t matter how much RAM you put in your SQL box. It’s not running. These queries don’t work in relational databases…

…So we did this for the same query yesterday in HDP and it took 20 minutes. This is amazing. We went from “it’s not possible” to “twenty minutes later it’s done.

Hortonworks Data Platform can store massive amounts of data from many sources and in many formats, and it unlocks data for these types of new discoveries. We’re partnering with our healthcare customers to help them make those breakthroughs.

The Healthcare Industry Uses Hortonworks Data Platform to Meet Common Data Challenges

At Hortonworks, we’ve seen our healthcare subscribers use big data to manage shifting reimbursement models, optimize labor and supply chain expenses and deliver improved care at a lower cost. They leverage advanced analytic applications to improve visibility of data already under management and also to optimize their data architectures for cost reduction.

I touched on some particular solutions in this post. In my next post in the series, I’ll go into more detail on some specific healthcare solutions with HDP and describe how those solutions are positively impacting both the financial wellbeing and clinical results of our subscribers.

Learn More About HDP for Healthcare

About the Author

Richard Proctor is the GM of Healthcare at Hortonworks, and is responsible for driving enhanced business value from data for the rapidly growing healthcare customer base. He has over 15 years in the industry focused on: supply chain optimization, enterprise intelligence, patient throughput, population health, capacity management, predictive modeling, interactive data exploration, quality improvement and geospatial visualization.

Tags:

Leave a Reply

Your email address will not be published. Required fields are marked *