Most common use cases for predictive analytics in healthcare

The State of Healthcare Analytics

Greg Nelson Analytics Competencies, Business of Healthcare, Healthcare Analytics, Our Thots

I guess like most people; I find myself a tad reflective this time of year. With another year quickly coming to a close, we will no doubt see the focus in healthcare once again turn to optimism as leaders prepare their organizations to take on the challenges of 2017 and try to predict what the new President and Congress will bring to healthcare “re-reform.”

Recently Modern Healthcare wrote on “The State of Predictive Analytics in U.S. Healthcare” where six healthcare leaders talked about the successes they have had in using predictive analytics to solve some of our most pressing challenges.

As one might predict (no pun intended), predicting hospital and emergency department readmission was a common theme for those interviewed. While none of the interviews went into much into the ‘hows’ of their analytics initiatives, a few highlights stood out for me.

Using Tertiary Data to Drive Predictive Models

Dr. Pamela Peele discussed the importance of social determinants in predicting outcomes, and as such, UPMC has turned to consumer and census data to calculate what is referred to as a deprivation index. Originally developed by Gopal K. Singh (2003), the index is a measure of the degree to which a patient is deprived of social resources based on the neighborhood in which they live. It should be noted that higher levels of deprivation have been associated with an increased risk of adverse health and health care outcomes. The index that Dr. Singh developed used 17 different markers of socioeconomic status and has been improved by several organizations since.

The beauty of this data is that the information is freely available and the index is well documented so that any health system could potentially use this simple algorithm to help guide interventions designed to improve access to care.

Understanding Patient’s Values for End of Life Care

One of the more novel approaches for utilizing predictive modeling was shared by Jeffrey Driver, CEO of The Risk Authority and Stanford Health Care’s Chief Risk Officer, where he described the gap that exists between a patient’s understanding of their prognosis and the physician’s. Stanford Health is developing a system to “read” a lifetime’s worth of medical information along with the patient’s personal documents. By interpolating the patient’s “values” they will compare that against the computable data to help the patient and better-informed decisions about their end of life care.

Data-Driven Culture

Another key theme coming out of these interviews was the degree to which healthcare organizations are adapting to a new culture of data-driven decision making. In our work, we are finding modest improvements in the extent to which organizations use and trust data as part of their implicit decision process. Our role in evangelizing and educating practitioners will no doubt continue to grow as confidence increases with trust in data and strides in data literacy programs. While demand for data continues to outpace the resources available, our hope is that continued progress in the area of analytical thinking and data literacy will support healthcare’s voracious appetite.

As Jim Dunn, the Chief Talent Officer at Parkland Health and Hospital System said, “We need to look at the ability to use data in decision-making as a core competency for onboarding and hiring leaders.”

My New Year’s Wish for Healthcare Analytics

Actionable Analytics Require Investment in People and process

Actionable Analytics Require Investment in People and process

My hope for 2017 is that every hospital will implement data literacy programs designed to build the nine competencies needed for data champions in healthcare.  I would also like to see that we will be able to fill an entire binder with use cases for advanced analytic success stories that go beyond the phrase “predictive analytics.”  We will look to data, quantitative methods, and the computing capability of 2017 to fundamentally change how we deliver care in the United States.