The 3 phases of creating an analytics-powered learning health system
Learning analytics isn’t much different than learning baseball, analytics pro Jason Burke says.
Five years ago, the healthcare industry was at the forefront of seeing the coming importance of Big Data, but back then, the electronic health record was the source of Big Data.
Now, that has been eclipsed by claims and genomic data, as organizations strive to create learning health systems, said Jason Burke, chief analytics officer at UNC Healthcare, during a session at HDM’s Healthcare Analytics Symposium in Chicago.
But what does it mean to have a learning health system? That’s a question the industry is still trying to understand, he noted. But there are building blocks that can help, particularly using data analytics and genomics to improve treatments and outcomes.
Using baseball as an example, Burke outlined a three-phase process of how an individual acquires skills. In baseball, there is a cognitive phase of rote memorization of holding a bat. Next comes the associative phase of the weight of the bat and how to hold it. Last comes the autonomous phase of operating the bat without really thinking about it.
These phases of creating skills can be replicated in a healthcare analytics environment by creating tools so clinicians can ask questions, such as what happens with patients at a certain level of acuity compared with general patients, and which treatments will work best with certain patients.
Don’t get stuck on the need for a lot of quality measures, Burke advised. “There are 9,000 quality measures. How will we know which ones actually matter? So, pick 10 or so that you can excel at.”
Burke also cautioned against centralizing analysts across an enterprise. These are the best subject matter experts an organization has, and they should be sprinkled in units across the organization, and they need not be data scientists.
That said, organizations do have to invest more heavily in data scientists, he added. Other subject matter experts are “reporters” who can explain who, what, when and where. But data scientists are “story tellers” who focus on the why and the how—they are basically investigative journalists.
Now, that has been eclipsed by claims and genomic data, as organizations strive to create learning health systems, said Jason Burke, chief analytics officer at UNC Healthcare, during a session at HDM’s Healthcare Analytics Symposium in Chicago.
But what does it mean to have a learning health system? That’s a question the industry is still trying to understand, he noted. But there are building blocks that can help, particularly using data analytics and genomics to improve treatments and outcomes.
Using baseball as an example, Burke outlined a three-phase process of how an individual acquires skills. In baseball, there is a cognitive phase of rote memorization of holding a bat. Next comes the associative phase of the weight of the bat and how to hold it. Last comes the autonomous phase of operating the bat without really thinking about it.These phases of creating skills can be replicated in a healthcare analytics environment by creating tools so clinicians can ask questions, such as what happens with patients at a certain level of acuity compared with general patients, and which treatments will work best with certain patients.
Don’t get stuck on the need for a lot of quality measures, Burke advised. “There are 9,000 quality measures. How will we know which ones actually matter? So, pick 10 or so that you can excel at.”
Burke also cautioned against centralizing analysts across an enterprise. These are the best subject matter experts an organization has, and they should be sprinkled in units across the organization, and they need not be data scientists.
That said, organizations do have to invest more heavily in data scientists, he added. Other subject matter experts are “reporters” who can explain who, what, when and where. But data scientists are “story tellers” who focus on the why and the how—they are basically investigative journalists.
More for you
Loading data for hdm_tax_topic #better-outcomes...