By: Jeff Deal, Program Chair, Predictive Analytics World Healthcare
In anticipation of his upcoming conference co-presentation at Predictive Analytics World for Healthcare Boston, Sept 27-Oct 1, 2015, we asked Chris Franciskovich, Data Scientist at OSF Healthcare System, a few questions about incorporating predictive analytics into healthcare. Catch a glimpse of his presentation, Preventing Readmissions and Reducing Costs with Predictive Analytics, and see what’s in store for the second annual PAW Healthcare conference in Boston.
Q: In your work with predictive analytics, what area of healthcare are you focused on?
A: My work crosses multiple areas of OSF’s integrated healthcare system. I’ve worked on projects focused on patient outcomes, ones that are more akin to traditional insurance projects, staffing efficiency projects and multiple quality improvement projects. Our predictive analytics projects are aligned to our strategic priorities, but are not constrained to specific operational functions.
Q: What outcomes do your models predict?
A: The upcoming talk in Boston will focus on the development and implementation efforts surrounding the 30 Day Readmission Risk model. It predicts the patient level risk of all-cause 30 day readmission and is designed to provide our clinicians and support staff the ability to proactively identify and mitigate patient risk.
Q: How does predictive analytics deliver value at your organization? What is one specific way in which it actively drives decisions or impacts operations?
A: We’re using predictive analytics to address a range of needs, from identifying our highest risk patients to assisting with more efficient staff scheduling. The 30 Day Readmission Model is currently being used to direct workflow activity in a variety of operational areas such as; inpatient case management, ambulatory care management, post discharge follow-up phone calls, outpatient palliative care and home care reporting/monitoring. Through the use of this model, we’re able to efficiently focus resources to those patients who are in the most need.
Q: Can you describe a successful result, such as the predictive lift of your model or the ROI of an analytics initiative?
A: Prior to the deployment of the 30 Day Readmission Risk model, I completed a comparative analysis between the then current approach, and the model. The then current approach achieved a cross validated AUC of 0.63 while the 30 Day Readmission Risk Model achieved a cross validated AUC of 0.76. The 30 Day Readmission Risk Model is also based upon data collected as part of the normal operational and clinical workflows of our organization. Thus, in addition to a significant increase in model performance, the new approach also provides the ability to approach our work in a much more efficient manner.
Q: What surprising discovery have you unearthed in your data?
A: Each project finds its gold, but I think the main discovery I’ve seen, from an organization level perspective, is the growing realization that we own the proverbial mine. Multiple projects have produced previously unknown insights from data we’ve been actively using for years or have proven that we can internally produce higher quality models than we can purchase. The success of the projects to date have solidified the organization’s realization that advanced analytics is major multiplier for the return on data technology and infrastructure related investments.
Q: What area of healthcare do you think have seen the greatest advances or ROI from the use of predictive analytics?
A: With the changes in payment approaches and proliferation of electronic medical records, it is an incredible time to work as a Data Scientist in healthcare. Historically predictive models have lived more in the insurance related areas of the industry, or have been simple and easy to calculate clinical tools. Both areas provided value in the past, but the availability of large amounts of structured and unstructured data, coupled with ample computing power and the financial motivations to compete on analytics makes for an amazing environment. I believe all areas of the industry are able to benefit from the appropriate use of predictive analytics.
Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World Healthcare in Boston.
A: The success, or failure, of a project is not solely tied to the performance of the model. You have to be able to translate the model into a meaningful story with which the business can relate. One of my favorite quotes is “Every act of communication is an act of translation.” As a Data Scientist, you must remember to function as a translator for both your business and its data.
Don't miss Chris’ co-presentation, Preventing Readmissions and Reducing Costs with Predictive Analytics, at PAW Healthcare on Monday, September 28, 2015 from 2:40 to 3:00pm. Click here to register for attendance,
By: Jeff Deal, Conference Chair, Predictive Analytics World Healthcare