This article was published by Mats Uddenfeldt, on his LinkedIn Pulse page [ https://www.linkedin.com/pulse/case-study-efficiently-converting-healthcare-data-mats-uddenfeldt ]. The article is republished here with the author’s permission.
Valence Health provides healthcare providers with customized solutions for value based care, helping them better manage their patient populations and accept financial responsibility for the quality of the care they provide. The company offers advisory services, health plan services and a suite of population health technology software as a service (SaaS) products that help their clients transition from a transaction-based approach to a value-based approach to healthcare. Headquartered in Chicago, Valence Health serves 85,000 physicians and 135 hospitals, helping them manage the health of 20 million patients nationwide.
The company’s rapid client growth and the increasing volume of data required to keep up with that growth were straining its existing technology infrastructure. “Our services have a voracious appetite for data. We use that data to inform decisions about improving both healthcare outcomes and operational processes. We knew if we continued to grow, we couldn’t sustain this,” explains Dan Blake, Valence Health Chief Technology Officer. “We had outgrown our Extract Transform and Load (ETL) infrastructure and knew we had to replace it.”
Valence Health looked at various alternative technologies. “We looked at large monolithic technologies like Informatica and point solutions like Syncsort but they did not give us the robustness and flexibility we needed in the long run,” says Blake. “We wanted something very open that would give us the flexibility to choose where to make investments over time. We were drawn to Hadoop and the capabilities those tools provided.”
Valence Health is using the MapR Converged Data Platform [ https://www.mapr.com/products/mapr-converged-data-platform ] to build a data lake that is the company’s main data repository. The company consumes 3,000 inbound data feeds with 45 different types of data including lab test results, patient vitals, prescriptions, immunizations, pharmacy benefits, claims and payment, and claims from doctors and hospitals.
“NFS was a very important feature for data ingestion. It is making our migration much easier,” says Blake. In the short term, Valence is transferring data back to the SQL server database as their portal and analytics expect that format. Once they get through the ETL transformation, they plan to transition from SQL to an HBase solution.
“We chose MapR [ https://www.mapr.com/why-hadoop/why-mapr ] because they were the easiest company to work with. They were the most receptive and answered questions quickly, honestly and efficiently. On the technology side, we’ve been very impressed with the overall ability to implement and integrate sequential data transformation.” Dan Blake, Chief Technology Officer, Valence Health
Valence Health is already seeing many benefits from its MapR solution including increased performance, better responsiveness to customers, higher quality data and a flexible platform to sustain their growth over time.
Growth requires new data architecture that can scale the business
Valence Health has been on a steep growth path over the last several years. “In the wake of the ACA’s implementation, more and more healthcare providers and organizations like Consumer Orientated and Operated Plans (CO-Ops) are taking on risk. As a result, we have tripled the size of our business in three years and expect do the same next year,” says Kevin Weinstein, Valence Health Chief Growth Officer. “Every year we’re more than doubling the amount of data we are processing. Having a robust data architecture is integral to our success.”
The company’s growth path is tied to client growth and the growth in the data infrastructure that those clients require. “We have to use technology to scale the business. We have to be able to manage more data with the same amount of people,” says Weinstein.
Performance gains increase customer satisfaction
The reliable and high performance of the MapR Platform enables Valence to be much more responsiveness to their customers. “In the past, if we received a feed with 20 million lab records, it would take 22 hours to process that data,” says Blake. “MapR can cut that cycle time down from 22 hours to 20 minutes [ https://www.mapr.com/company/press-releases/valence-health-dramatically-improves-data-ingestion-performance-and ] . And it’s running on much less hardware.”
“MapR gives us the resource efficiency, speed and flexibility to make a huge difference in customer satisfaction. As soon as the data hits our system it’s pushed all the way through,” he says. “It gives our customers much faster feedback about what’s going on with the population they are trying to manage.”
Flexibility serves customers faster
Valence Health is also now able to answer customer requests that were very difficult to answer in the past. “It allows us to do things we could not do in our old world,” explains Blake. “For example, a customer might call and say: ‘I sent you an incorrect file three months ago and I need you to take that file out.’ That’s not something you can do in a normal ETL system on top of a relational database. It could take 3-4 weeks to get that data deleted,” he says. “But with MapR, that is naturally supported, we can just roll it back and take that file out.”
This ease of administration and maintenance means that the company can focus more resources on their core business. “I can spend less on outsourced resources and instead spend money on adding new features, analytics or visualization capabilities or acquiring new types of data. We can do things that truly matter to our customers,” says Blake.
Enriching the data lake with new data sources enhances data quality
The MapR Platform also makes it much easier for Valence Health to enrich their data lake with new data sources. “It’s not just the volume of data, we’re looking to integrate new types of data like socioeconomic and demographic information, or immunization records. With our old architecture it was painful to do so,” says Blake. “Our data scientists are looking at new sources of data. The data can tell you things you don’t even know about. If we can augment our data, we can build new types of analytics that allow our client and our company to successfully invest in areas we have not been in before.”
Data acquisition and integration capabilities enable differentiated services
Valence Health believes that their new data acquisition and integration capabilities will give them a leg up over their competitors. “Other startups are selling software solutions in spaces we operate in,” explains Blake. “But the hard part is getting the data into the system and into formats where it can be truly useful. Our twenty years of hands-on practical experience in working with provider organizations that have taken on all sorts of risk-arrangements coupled with our effective and efficient infrastructure to get data flowing and keep it flowing is hugely powerful. The data acquisition infrastructure is very important to our ongoing success and to our customers.”
Want to learn more about Big Data for the Healthcare industry? Please check out the links below:
How Health Care IT Diagnoses Data Pain Points, CIO Insight – http://www.cioinsight.com/case-studies/how-health-care-it-diagnoses-data-pain-points.html
Stepping Up to the Life Science Storage System Challenge, HPC wire – http://www.hpcwire.com/2015/10/05/stepping-up-to-the-life-science-storage-system-challenge/
Health Care Emerges as Hadoop Use Case, Datanami – http://www.datanami.com/2015/10/08/health-care-emerges-as-hadoop-use-case/
Big Data and Apache Hadoop for Healthcare and Life Sciences, MapR – https://www.mapr.com/solutions/industry/big-data-and-apache-hadoop-healthcare-and-life-sciences
Join our more than 700+ paying customers and discover why MapR [ https://www.mapr.com/ ] is the clear market leader for production ready Big Data applications by reading about the Top 10 Reasons Customers Choose MapR [ https://www.mapr.com/top-ten-reasons ].
This article was published by Mats Uddenfeldt, on his LinkedIn Pulse page [ https://www.linkedin.com/pulse/case-study-efficiently-converting-healthcare-data-mats-uddenfeldt ] . The article is republished here with the author’s permission.