agsandrew - Fotolia

Manage Learn to apply best practices and optimize your operations.

How to speed up analytics in healthcare

Attendees at the TDWI conference in Boston explain how they are speeding up analytics in healthcare.

Healthcare, more than many other industries, needs quick data analyses. When a patient has a check-up or is about to be discharged from a hospital, doctors may have just a couple of minutes to explain their instructions to the patient. As the industry comes to rely more on analytics, providers are looking for ways to speed things up.

For John Poonnen, vice president of product management at Community Care of North Carolina (CCNC), part of the answer to speeding up analytics in healthcare is data virtualization. He explained at the TDWI Executive Summit in Boston that the concept involves making data available for analysis in a system without actually moving the data to that system or to a common data warehouse. His group uses tools from Informatica and Cisco to connect various data systems and share information remotely.

Currently, CCNC is applying analytics to patient admission, discharge and transfer records to cross-reference them with billing and claims data. The goal is to identify patients who are likely to be readmitted after leaving the hospital. One provision of the Affordable Care Act will penalize providers for avoidable readmissions, which is one reason CCNC wants to reduce that number.

Poonnen said that virtualizing the data allows providers to quickly access it and make decisions.

"[Clinicians] don't want to wait a long time for data to be brought into the data warehouse," he said. "They want to see the data and see what they're missing. Not everything can be brought into the data warehouse with the level of latency needed to take some action."

Still, this doesn't mean that providers can get away with poor data governance. Failing to assign ownership of data stores to specific departments, provide for required accessibility and maintain consistency across data sets can quickly erode the effectiveness of a data virtualization project.

"Data virtualization is not the Wild West," Poonnen said. "It doesn't mean you can get away without data governance."

For Andrew Proctor, senior director of business intelligence in the Medical Operations Division at the Cleveland Clinic, the answer is data visualization software. He said during his presentation at the conference that Cleveland Clinic uses SAP Xcelsius and Tableau software to develop visualizations for managers to examine patient length-of-stay information. This allowed them to identify patients who were likely to stay longer than expected and develop strategies for dealing with them.

Proctor said there are many visualization tools available that are simple to use and enable business users to create their own graphs. This eliminates some of the need for IT or analysts to be involved. Often users will develop reports for themselves that could be rolled out throughout the organization. Getting more people involved reduces development time.

However, providers need to choose their tools wisely. Cleveland Clinic began relying on data-driven reports in 2002, when most of them were done in Microsoft Excel. Since that time the provider has adopted various database systems along with reporting and dashboarding tools. But Proctor said providers need to be more thoughtful about the tools they adopt and how they are integrated.

"You probably own every product out there," he said. "They're not integrated, they're not governed. We can't afford to do that anymore."

Ed Burns is site editor of SearchBusinessAnalytics. Email him at and follow him on Twitter: @EdBurnsTT.

Next Steps

How providers can get around HIPAA to do analytics in healthcare

Why health analytics and quality improvement teams must work together

Providers need to do more to adoption analytics technology

Hospitals and healthcare systems begin analytics implementation

Dig Deeper on Predictive analytics

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.