This is the final piece of our three-part series on using predictive models in healthcare. Be sure to check out parts one and two for more information on how predictive modeling can be used to attract and identify patients.
Predictive modeling allows healthcare marketers an inside view into their big data. With patient modeling, marketing departments have the ability to score potential patients based on a set of criteria, by identifying and targeting those who are most likely to respond. Marketers can then tailor communications to those individuals.
The use of predictive modeling, or propensity modeling, helps marketing departments better utilize their marketing dollars and cut down on inefficient marketing spend. Evariant has incorporated this into their engagement platform, allowing their users to make smarter marketing decisions.
Predictive Models and Service Lines
These models are able to identify a specific subset of patients. A core set of models start from the service line and sub-service line levels. The initial sets of service lines are derived from a combination of historical ICD-9 and DRG codes as well as procedure codes and proprietary information unique to each client. In addition, depending on business and campaign rules, many clients choose to blend sub-service lines into targeted market specific groups.
Examples of Predictive Models
The Female Health Model
One example of this is the female health model. A healthcare system houses the majority of their female oncological services using a centralized service delivery system. Due to the organizational structure, this healthcare system would be best served by a blended female health model.
This is achieved by combining multiple sub-service lines (mammography screening, breast cancer, and female reproductive cancer) into one overall propensity model (for both patient and consumer prospects).
Orthopedics can also find value in utilizing this model. Healthcare systems use several orthopedic marketing models specifically to target younger customers, both patients and non-patients that were often involved in athletics.
A critical issue in the front-end analytics is to perform group-difference and other analyses to determine similarities and differences between and within groups. In the case of this example, analyses showed that younger orthopedic patients were statistically different from older patients, and that sports-related orthopedic injuries could be statistically separated from non-sports related injuries
Understanding this information provides great insight for healthcare marketers and helps to realize a positive ROI for healthcare marketing departments. Models are an effective way to gather, visualize and identify where the most likely potential patients are and how to communicate with them.