PAW’s Predictions on the Not So Easy World of Analytics
By Sarianne Gruber
Welcome to 2018! This year Predictive Analytics World (PAW) wants to know in the world of analytics and healthcare “Is there an ‘Easy Button’ for Analytics?” PAW moderator Jeff Deal, VP Operations for Elder Research recalled the slogan “that was easy” from the popular commercial expressing how easy it was to buy office supplies. “Easy in analytics takes on a whole new meaning. It is never going to be like buying office supplies or installing Microsoft Office. Has bringing analytics to healthcare has it gotten any easier than 5 to 10 years ago? Are we making progress in automating the analytic industry?” posed Deal to three distinguished panelists. Addressing these not so easy questions were panelists: John Elder, Ph.D., Founder and Chairman, Elder Research, Inc., Ken Yale, DDS, Chief Analytics Officer, Delta Dental of California and Pamela Peele, Ph.D., Chief Analytics Officer UPMC Health Plan & UPMC Enterprises. Note that the material presented below has been gently edited.
Where the Future Investment lies in Analytics
Automated Data – Different sources of health and medical data have varying degrees of accessibility. In a claims payment system much of the data is fully automated and very adaptable. This type of system can easily be designed to send alerts when something is over-utilized. “In a situation where you are constricted or constrained because you have to use EHRs, you need to figure out how to normalize and standardize the data, which is a real disadvantage,” Dr. Yale affirmed. A long-term handicap has been our acting legislation, logged with at least five years of past testimony, and when passed becomes codified into law – as in case of the Affordable Care Act with electronic medical records. Inadvertently, this has led to restricting data to a certain shape and form. It eliminates the means for innovation and doesn’t necessarily comply to the way hospitals are run. “If you are in a system where you are stuck in a regulatory environment where you have to report to regulatory or meet CMS requirements you are at a huge disadvantage. And, yes it very difficult both timely and costly. But I believe we are going towards the digitalization of everything so that we can actually work outside that system and get a lot better results. There is an easy button but we are not there. I think in the next 3 years we will see some very interesting things,” predicts Yale.
Infrastructure – “Investment needs to be in making the data fit for consumption. The University of Pittsburgh Medical Center invested 2 billion dollars in analytic infrastructure and operations. That’s no easy button” conceded Dr. Peele. In her opinion automating analytics is like putting the cart before the horse, especially when most organization don’t have their data together. She thinks attention must first be first in establishing quality data, before hiring data scientists or purchasing “fancy” software.
Customer Behavior – The next overlay to clinical and claims data is Customer Behavioral Data. More and more we want to know things like what type of food patients are purchasing or are they drinking too much. There are several vendors that collect consumer data, and Delta Dental selected Axciom. Its data dictionary contains 1800 health-related variables from cooking behavior to shopping behavior. “We had an algorithm that could predict my health score and a number of other variables based on predictive analytics using claims data and some clinical data. When we combined the model with some consumer data, it was amazing what can be learned from what you buy and what you eat. The consumer data is very, very specific. It is like a crystal ball. Combing this rich data with healthcare data at the point in time when you saw your doctor or when you went to the hospital, the results become extremely robust and very useful,” said Dr. Yale with encouragement.
Cultural Change – “It doesn’t matter how ‘scary smarty pants’ we are, how much computational power we have or how good our algorithms are for artificial intelligence. We must ingest what we are learning analytically and feed the narrative to the organization,” exclaimed Dr. Peele. She contends that if the organization can’t or won’t consume analytics, the organization doesn’t change. And even if it is consumed, but the organization isn’t changing based on what we are surfacing, the value of what we are doing is zero. The most effective change is taking what we learn through our intelligence and delivering results to the industry leaders. Dr. Yale then interjected, “A lack of change is an impediment to finding that easy button.”
Leadership – Moderator Jeff Deal joined in the discussion from the consulting perspective. He contends that without strong organizational buy-in and strong leadership, it doesn’t matter what easy buttons are out there. You are not going to get the buy-in from the people throughout the organization unless you have buy-in at the top, where the budget is determined and most importantly where cultural change derives. The expectation how you are going to succeed in an organization funnels through how the organization operates. Dr. Elder recommended highly Jeff Deal’s and Gerhard Pilcher’s book entitled Mining Your Own Business to help learn what issues need to be addressed to help decision makers get over the hurdle of accepting analytics. “Healthcare is a tough one because you have privacy issues with data, headstrong physician end-users and pushback getting people to change how they are doing everything, all of which is a tough sell. And yet, it is possible,” stated Elder optimistically.
Learn more about Predictive Analytics World and their upcoming conferences.
Find out more about the Predictive Analytics World 2017 Fall conference in New York City.