ORLANDO, FL — More radiology practices and departments are beginning to see the promise and opportunity in business intelligence and data analytics tools. But implementing this software is no easy task. A panel of vendor experts speaking at SIIM 2012 Friday offered some sage advice to users considering a BI project.
• Know your data, said Matt McLenon, CEO of Softek Inc. Most business intelligence applications rely on solid HL7 data, but HL7 data isn’t always easy to understand. It helps to have someone on the radiology staff — not just the IT staff — familiar with what data is collected and how it’s structured. Are the elements missing or mislabeled? Check the data first.
• Get all the stakeholders involved, McLenon advised. For example, make sure the radiology chief articulates what data he or she wants to access from a business intelligence dashboard. If the chief wants to know the productivity level of all the radiologists, but doesn’t want the rest of the staff to have that access, that desire should be laid out before the implementation begins. It’s easier to set up a system to meet those needs before, rather than after it’s live.
John DeLong, VP of marketing for Medicalis, echoed this sentiment. Secure senior level buy-in, he said, because the project will require internal staff and resources, which must be factored in when considering the value of such a system.
• Understand what the end users are expecting. “People look at BI as a silver bullet that will just get their questions answered,” McLenon said. But the radiology group needs to know how they plan to act on that data. End users need to be a part of the implementation, and work with their vendors to articulate their expectations.
Various users in the organization are going to have differing views on business analytics, DeLong said. Radiologists will want to view data relevant to their clinical work flow, while department heads might be interested in business decision-making tools such as longitudinal dashboards and notifications if the department is above or below a set threshold. A business intelligence tool can account for those various layers and requirements, and understanding each users’ needs will ease the implementation.
• Define how you will evaluate the success of the system, said Rob Fleming, product management director, diagnostics, for Nuance Healthcare, who spoke in reference to the data mining capabilities of speech recognition tools. Most often, radiologists will define success through turnaround times or a reduction in transcription costs, but these systems are chock full of data that can be mined and used, he said.
“What are you trying to achieve so you make sure you configure your systems to capture and report that data?” he asked.
The radiology group has to ask the tough question, DeLong said: Can you extract business and clinical value from your data? Answering this question first will ground the project, he said, and guide the implementation. “Too often BI projects start with, ‘Let’s just organize the data,’ but we try to determine the success criteria for any BI project.”