Introduction to Ranking Models - Text Mining Techniques for Healthcare Provider Quality Determination - page 1

Chapter 1
Introduction to Ranking Models
IntroductIon
In this text, we will discuss how patient risk adjustment models are defined, their shortcomings, and their
benefits. We will also provide some innovative methods to improve on the currently used risk adjustment
methods. Risk adjustment is necessary to examine the differences between healthcare providers. Sup-
pose, for example, that hospital A has a higher mortality rate compared to hospital B. Without looking
at whether or not hospital A treats sicker patients at higher risk for mortality compared to hospital B,
it is impossible to determine whether hospital A has better or worse outcomes compared to hospital B
for treating any one patient. According to the Society of Thoracic Surgeons, risk adjustment is a way
of “leveling the playing field” to adjust for differences in risk among specific patients. Risk adjustment
makes it possible to compare different healthcare providers in terms of quality.
However, because patients are so different from each other, it is very difficult to define the “typical”
patient. How does a patient with severe heart disease compare to a patient with kidney failure? Do they
have similar risk, or does the heart patient have a higher risk than the kidney patient? Without finding
some way to determine “comparable risk”, there are just too many combinations of patient co-morbidities
to compare two providers on a nearly identical patient.