Taking It to the Cloud: How the American College of Radiology Boosted Its Operations
With an eye toward business intelligence, ACR leveraged cloud resources on its way to building a data lake. The technology was important—and so too was making its capabilities clear to staff and stakeholders.
Associations often have to manage systems across departments with differing needs. When those systems are at odds, it can create productivity and infrastructure challenges.
This was the issue facing the American College of Radiology, a medical society with nearly 40,000 members. Ultimately, the solution came in the form of the cloud. Shree Periakaruppan, ACR’s director of data engineering and analytics, said that the association moved its data infrastructure to the cloud in the form of a “data lake,” a less-structured approach to data management that grants more flexibility for the association’s various teams, leading to benefits in areas such as business intelligence.
“I think for any organization that is considering moving to the cloud, there needs to be a vision,” Periakaruppan said. “The why is extremely important—why you’re moving to the cloud—and then the alignment is even more important once the why is identified.”
In the case of ACR’s “why,” the organization was running into performance issues with its existing reporting approach, which relied on Tableau. Trying to make these databases work within a business-intelligence infrastructure proved challenging and required stopgap solutions, such as nightly synchronization of data into a reporting server. ACR also needed to be HIPAA-compliant, a common factor for healthcare associations.
The Transformation Strategy
ACR landed on a data lake approach, storing data of all types in an accessible raw format. A data lake also allows for smaller use cases to be accessed from a larger pool of data, rather than taking a more structured database approach, which was leading to performance issues internally.
“There was little need for curating all the new datasets, and hence it was efficient for us to only structure or curate datasets that needed to be reported,” Periakaruppan said. “The rest of the data was just stored in the lake until there was a need to build intelligence on top of it.”
After testing a variety of solutions, including some on-premises solutions, ACR decided upon Amazon Web Services.
“Most of the on-prem databases were either expensive or were not flexible enough to handle our frequently changing data storage and processing needs,” she said.
The association worked through these issues and built a data pipeline so data could be consumed and organized through a data warehouse (a system for organizing data from disparate sources), at which point the data could be pulled into Tableau more effectively.
(Periakaruppan spoke at a recent AWS event for nonprofits on this transformation effort.)
Beyond improving the data pipeline for business intelligence, the move to the cloud had other benefits, including the ability to leverage speedy procurement of cloud resources.
“Once the discovery and definition period is complete, we can design and quickly prototype in AWS. This helps us validate the solution quickly and iterate on the design and development at a faster rate,” Periakaruppan said.
Selling the Data Lake
The data lake was an elegant solution, but it still needed buy-in from the staff. Periakaruppan and her team put in the work to effectively communicate the benefits of the data lake infrastructure, including training employees and integrating other types of data into reporting dashboards, such as data from Google Analytics.
The result is that ACR has made major strides in its digital transformation work, to the point where nontechnical employees feel more comfortable with it.
“The adaptation has been slow, and it might take a while for all business units to embrace the data lake for advanced analytics, but we are moving in the right direction,” she said.
While Periakaruppan said that ACR has made major strides in improving its infrastructure, she said that there’s still plenty of room to grow. She’s working to create a cloud operating model, which will help to ensure that the organization is properly leveraging its resources.
For example, the association will build a process for maintaining untagged resources.
“We need to tag appropriately, so if there are resources that are not being tagged, they would automatically be shut off, or they would not be usable, so that people make sure that they tag it appropriately,” she said.
Ultimately, though, the association has found an important way to better help it meet its broader goals. And not just for its staff: Using a data lake allowed ACR to create dashboards and datasets that help improve interactions with members.
“Moving all our data to the data lake and the capabilities it has brought with it has helped us make huge strides toward our core purpose,” she said.
(iSergey/iStock/Getty Images Plus)