Experts take opposite sides in use of ‘dirty data’ in analytics
Debate continues on the need for high-quality information to achieve usable results, says Charlton Park.
To clean data, or not to clean data—that is the question. Generating clean, high quality data is often considered to be the holy grail of analytics. However, John Showalter, MD, chief health information officer at the University of Mississippi Medical Center, believes that perfect can be the enemy of good.
According to Showalter, data “doesn’t need to be clean to be trusted.” In fact, he contends that cleaning data—an expensive, time-consuming process—does not improve analytics. “Send it dirty—whatever the lowest amount of effort is what I encourage you to do,” Showalter told an audience at HDM’s Healthcare Analytics Conference held last week in Chicago.
At the same time, he believes that the data needs to be “timely, accurate and complete” to be trustworthy.
“We need to have that trust that we’re looking at data that’s actually really true, and then we have to have faith in the analytics—which is where most people fail,” Showalter said. “If doctors don’t have faith in the analytics, then they won’t act.”
Ultimately, Showalter argued that it’s not about the right data—it’s about the right engagement with stakeholders. “There is no right or wrong data. It’s about who’s at the table with you.” The bottom line: if all the analysis creates actionable insights that don’t result in action on the part of providers, then there is no way that better health outcomes can be achieved through medical interventions, he said.
But purity of data is critical, contends Charlton Park, chief analytics officer at the University of Utah Health Care. Park contends that the integrity of data is paramount, especially given that dirty data is so pervasive in healthcare. According to Park, the successful use of analytics in healthcare requires data that is transparent, accurate, actionable and engaging.
Transparency is the product of “building a culture based on openness to data,” he said. “Accuracy is very, very important,” Park added. “Data is only as good as it is accurate.” He also made the case that data must be “detailed enough to be actionable” and that analytics tools should engage doctors and be understandable to them.
“Without the physicians, nothing’s going to happen,” concluded Park.
According to Showalter, data “doesn’t need to be clean to be trusted.” In fact, he contends that cleaning data—an expensive, time-consuming process—does not improve analytics. “Send it dirty—whatever the lowest amount of effort is what I encourage you to do,” Showalter told an audience at HDM’s Healthcare Analytics Conference held last week in Chicago.
At the same time, he believes that the data needs to be “timely, accurate and complete” to be trustworthy.
“We need to have that trust that we’re looking at data that’s actually really true, and then we have to have faith in the analytics—which is where most people fail,” Showalter said. “If doctors don’t have faith in the analytics, then they won’t act.”
Ultimately, Showalter argued that it’s not about the right data—it’s about the right engagement with stakeholders. “There is no right or wrong data. It’s about who’s at the table with you.” The bottom line: if all the analysis creates actionable insights that don’t result in action on the part of providers, then there is no way that better health outcomes can be achieved through medical interventions, he said.But purity of data is critical, contends Charlton Park, chief analytics officer at the University of Utah Health Care. Park contends that the integrity of data is paramount, especially given that dirty data is so pervasive in healthcare. According to Park, the successful use of analytics in healthcare requires data that is transparent, accurate, actionable and engaging.
Transparency is the product of “building a culture based on openness to data,” he said. “Accuracy is very, very important,” Park added. “Data is only as good as it is accurate.” He also made the case that data must be “detailed enough to be actionable” and that analytics tools should engage doctors and be understandable to them.
“Without the physicians, nothing’s going to happen,” concluded Park.
More for you
Loading data for hdm_tax_topic #better-outcomes...