Q&A: Advanced Analytics Crucial to Healthcare Evolution (Part 1 of 2)
In healthcare's ongoing transformation, analytics -- in particular prescriptive analytics -- is playing an important and growing role, explains Meg Aranow, an expert in healthcare and BI.
- By Linda L. Briggs
- March 3, 2015
"Prescriptive analytics is going to play a major role in managing healthcare priorities," according to Meg Aranow, senior director of research and insights at The Advisory Board Company, a membership-based global research and consulting firm focused on healthcare.
Aranow has 25 years of experience in healthcare, with expertise in business intelligence and analytics, IT strategy and planning, clinical information systems, electronic medical records, and vendor assessments. Prior to joining The Advisory Board Company, she served as VP and CIO of Boston Medical Center and held leadership positions at Partners Healthcare and Brigham and Women's Hospital. In 2009, she was appointed by the Massachusetts governor to the Healthcare IT Council for the Commonwealth of Massachusetts, which oversees the state's development of a regional extension center and health information exchange.
In this two-part interview with TDWI, she outlines the role that analytics is playing in healthcare -- and will play in the future. "In the healthcare transformation taking place," Aranow says, "we think that prescriptive analytics will play a big role in the years to come."
BI This Week: Can you tell us a little bit about your firm, The Advisory Board Company, and its role in healthcare?
Meg Aranow: We're a 35-year-old company founded in 1979. We provide best-practices research to member organizations. Our members subscribe to access our research, whether it's written research or consultation with experts who produce syndicated research. We are dedicated specifically to healthcare and healthcare strategies. Note that when I talk about healthcare in this interview, I am talking primarily about the membership of the Health Care IT Advisor program within The Advisory Board, which is mainly made up of those that provide direct care -- usually a hospital or a clinic -- as well as the software and solution providers that service those people. That means that when I'm talking about analytics specifically, I'm not talking about Big Pharma [or insurance companies], for example -- that's outside the umbrella of what my particular program addresses.
In a recent TDWI Webinar, How Prescriptive Analytics Will Shape the Future, you described three levels of BI maturity in healthcare, "with each level more difficult and more advantageous than the last." What are the three levels and how do they differ in regards to healthcare?
The three levels of BI functionality that we discussed are descriptive analytics, predictive analytics, and prescriptive analytics. We've found the three levels to be a good construct as we're talking to members about healthcare. It makes for a good shared lexicon and a level playing field, and they're reasonably well-known terms within the business intelligence space.
As you noted, as we move through the phases, we are increasing the complexity of what's involved, and also likely increasing the value of what's being returned to your organization.
With descriptive, we're talking primarily about a retrospective view of data. We're looking at the past and counting events that happened in the past. For example, we might be counting admissions at a hospital or the average number of prescriptions for a patient.
With predictive, we're throwing that forward and saying, based on that history, what might we expect to see in the future? We're using data to predict, and therefore prepare for, those things coming toward us in the future.
When we get to prescriptive analytics, it's also future-facing. It has more to do with interventions we might make -- changes that we may want to make today so that we can bend the trajectory of the future based on what our predictive analytics tell us. The goal is a more optimal outcome.
A very common example in healthcare these days is readmission following a stay at an acute care facility. Medicare and Medicaid have -- as part of the transformation taking place in healthcare -- begun to reimburse or penalize hospital organizations for the quality of healthcare they are rendering. As one indication of quality, they are looking at unplanned re-admissions -- when someone comes back to the hospital unexpectedly post-discharge. When that happens, perhaps the person shouldn't have been discharged as soon as they were, or perhaps there were unforeseen complications that sent them back to the hospital.
If we play out that idea of hospitals focusing on unplanned readmissions in our analytics models, we would use descriptive analytics to go back over the last three to six months and see where our unplanned readmissions are. How many did we have, in what types of patients, under the care of which doctors, or of what ages -- in short, what did they have in common? We try to understand what's happening through the use of descriptive analytics, looking at things retrospectively.
In the predictive phase, we would then use that data to try to understand what we might expect in the future. Would we expect these trends to continue? How many unplanned re-admissions might we see in the next 30 days?
With prescriptive analytics, finally, we might be thinking about what kind of interventions to take with a patient who we predict might be an unplanned readmission. What kind of intervention might we take to prevent that readmission from happening?
Keeping with that model, perhaps what I've learned through descriptive analytics is that I have a lot of patients who had cardiac procedures who were over the age of 80, who unexpectedly came back to the hospital as unplanned readmissions. I would expect that, if nothing changes, that would continue to be true over the next 30 days. I can see who I have scheduled for those procedures who fit in that category -- cardiac patients over the age of 80 and so forth -- who might be at risk for an unplanned readmission. Using prescriptive analytics, I might determine what kinds of interventions I could take with them to prevent those unplanned readmissions from occurring.
As an example, by using optimization routines, maybe I've found patterns to indicate that I would have had a more optimal outcome if I were able to better place those patients in extended care facilities that were closer to family members who could interact with them on a social basis, or if the patient were assigned to someone with a specialty that could deal with some of the relevant co-morbidities.
So you can see how I might use prescriptive analytics to understand what actions I should take today to change predictive outcomes tomorrow.
Looking at the healthcare sector that you focus on, which is providers, where are most provider organizations along the curve of analytics you've described?
Remember that these are cumulative stages -- it's not that one replaces the other, but rather, organizations can use one, two, or all three types of analytics.
That said, most organizations with whom we speak are very well entrenched in and are practicing descriptive analytics -- that retrospective view of the data. They are also beginning to make some progress on predictive analytics. There are very few organizations that are doing anything substantial with prescriptive analytics yet.
Healthcare is sometimes said to have been slow in getting started with BI and analytics. Given that, how would you rate its progress now?
You're right, when we compare ourselves to some other industries, we've been a little bit slow on the uptake. However, we have every indication from our member surveys and conversations that CIOs in healthcare are now quite focused on analytics. We see that reflected in surveys in which we ask our members -- over 300 healthcare organizations -- to identify their top 10 issues of interest and concern; 2015 will be the fifth year in which BI and analytics rated in the top five issues.
We also see that attention is backed by funding -- our members are making purchases in the BI software space, and also looking to acquire talent and staff. In a survey of healthcare leaders conducted in the fall of 2014 interest in BI was rated second only to tele-health. How quickly will they catch up? I can't say for sure, but their attention is in the right place now.
I would add that we are doing our best to raise the profile of prescriptive analytics through events such as the recent TDWI Webinar I participated in. We do see that prescriptive analytics is going to play a major role in being able to manage healthcare priorities -- not only the challenges we have today but those in the future. In the healthcare transformation taking place, we think that prescriptive analytics will play a big role in the years to come.