Thursday, May 26, 2016
The population health management world is filled with wonderful dashboards and data visualization tools - aka "shiny objects". Many of these tools are becoming core components of healthcare analytics. I have often been asked over the years by people looking for something to help them manage a patient population which tools made the most sense for their situation.I have also read countless RFPs that tried to rank tools based on certain features. On the surface, all of that makes sense. However, after being in the industry for the better part of two decades, I have learned that the tools are simply a shiny object much like a flashy car. To provide a simple analogy, deciding between two visualization tools is like deciding between a Maserati and a Ferrari. Both are fabulous cars. However, what is more important is the gasoline, because if you can only get high octane gasoline for the Ferrari, the choice between the two is simple.
So what is the high octane gasoline? If you are guessing data, you are wrong. Data cannot fuel an analytics tool with meaningful analyses any more than crude oil can fuel a Ferrari. The oil must be refined and so must the data. In healthcare analytics, this refinement involves many, many steps. Some people think this is data scrubbing, i.e. cleaning up invalid codes. That is one of the steps, but it if far more than that. If you are bringing in data from 40, 50, even 100+ different sources, you want the user not to have to worry about which source the data came from. You want it to all look the same. Then the big step, the member/patient match. You have to aggregate the data from all of those disparate data sources around each member. More importantly, you have to run rules against that data...thousands of them in fact, in order to create any content that can be useful in healthcare analytics.
For example, raw data will indicate a diagnosis of diabetes. Without a rules engine, everyone with a diabetes code in their record will be considered a diabetic. Nothing could be further from the truth. Predicting diabetics out of a population involves very complicated rules that must be continually refined. Determining chronic conditions is simply one of many rules sets that need to be applied. Once these rules have been applied, you now have something tangible to analyze, and maybe even put that analysis in a shiny object like a dashboard or visualization format as a final step.
If you pick a population health analytics vendor that is not adept at data transformation, the heavy lifting of refining raw data into meaningful information, all the analytics performed on the information will be inaccurate at best. You might have great visualizations from the shiny object you bought, but it will be great visualizations of incorrect information. The lesson here is do not assume that everyone can handle the hard work of data transformation. When I go to HIMSS and talk to vendors about their applications, I always start the conversation by saying I am not interested in the shiny object on display. I assume that looks great, as that's the easy part. I am far more interested in what is behind the shiny object. I would recommend that anyone else looking into a healthcare analytics partner do the same.
To learn more about Deerwalk's Population Health Managment portfolio of software and service solutions, please visit www.deerwalk.com