juanjo tugores - Fotolia

Big data analytics architecture requires integration push

Organizations looking to analyze big data typically have to pull it together from various systems, making data integration a fundamental component of a big data analytics platform.

For most businesses that have embraced big data analytics, the IT world in which the enterprise data warehouse was the one place where companies stored any data worth analyzing is a thing of the past. Today, such organizations typically possess a variety of analytical data stores, still including a data warehouse but also the likes of Hadoop clusters, NoSQL databases and department-level data discovery systems.

There's potential business value in combining data sets from different platforms for analysis and data visualization -- uncovering that value is one of the primary goals of big data applications. But before analytics teams can start their work, organizations often need to develop an integrated big data analytics architecture so data scientists and other analysts can access all the information they're looking to explore.

For example, merging data pulled from a customer relationship management system with call center notes on interactions with customers could give a company a better understanding of issues with its products and services. And adding social media data to the analytics mix could provide a fuller view of customer sentiment. There are multiple ways to go about blending such data so it can be analyzed and visualized -- the technology options include conventional data integration software, pre-built connectors between systems and integration functionality built into data discovery and visualization tools.

Advocate Health Care, which operates 12 hospitals and dozens of other medical facilities in Illinois, turned to a cloud-based integration platform to link data from various electronic health record (EHR) systems it runs internally, plus outside info such as claims data from insurers, pharmacy records and material from government databases. Advocate came together through a string of mergers and acquisitions, and its EHR systems were largely disconnected from one another before the start of the big data integration initiative -- but standardizing on a common system would have cost billions of dollars, said Tina Esposito, vice president of the company's center for health information services

Instead, the HealtheIntent platform, developed by health IT vendor Cerner Corp., creates registries containing standardized information about patients with similar conditions, giving doctors graphical views of how individual patients compare to both a broader group and predefined standards on care metrics. A separate EHR module places a color-coded readmission risk score, refreshed every two hours, at the top of patient records to show care teams how likely people are to require a return to the hospital within 30 days.

That kind of information is crucial to Advocate's efforts to adopt an accountable care payment model, which puts healthcare providers on the hook financially for the health of their patients. It would be impossible to operate under the new reimbursement model without the unified view of patient data provided by the big data analytics architecture, according to Esposito. "Unless you have all of this information pooled on a patient," she said, "you can't deliver targeted interventions."

Next Steps

BI architecture has catching up to do to keep up with new analytic demands

Keep these things in mind when putting together your big data environment

Big data architecture needs planning to get to analytics value

Intel advances new big data analytics platform to expedite development

Dig Deeper on Big data analytics