It All Started With a Simple Request (2019-US-30MP-185)
Aug 27, 2019 12:44 PM
| Last Modified: Oct 17, 2019 2:49 PM
Jim Grayson, Professor, Augusta University
It all started with this request: "Can you help us determine why there are so many no-shows for our doctor’s appointments?" In response, using JMP predicitve modeling and after many iterations, we developed a beautiful predictive model to identify the characteristics of the no-shows. Although we were able to use our expertise to answer the initial question, once we “solved the problem” we realized that the true need was not to develop a wonderful model. Instead, we needed to use what we had learned from our model to create interventions that would lead us to eventually understanding the true goal: getting patients to the clinic for their visits!
We pivoted the project into phase two and developed a simple designed experiment to study potential interventions to address the insights we had learned from the initial predictive model on the no-show population. This study completed in April 2019, and results showed very significant improvements for two of the three interventions.
We are reminded that there is a beginning, a middle and an end. We get the most excited about the middle – that is the modeling. But what makes or breaks us are the beginning and end – the beginning is understanding the true goal, and the end is when we deploy the results that meet the true desired outcome of the project.
In this talk we will discuss this entire process, from starting this project with a primarily analytic goal to pivoting to a better business-focused goal that went full cycle from problem statement to deployment.