Risk Model Selector

Using WebMD Ignite’s own data insights, the Risk Model Selector suggests the best consumer risk or patient risk models to select the right audience to meet your campaign's objective. You can choose to launch the Risk Model Selector workflow when you apply either the Consumer Risk criterion or the Patient Risk criterion to an audience in Audience Insights.

Using what you know about the audience you are looking to target for your campaign, you choose one of the following ways to arrive at a list of suggested models specific to your campaign:

• For a service line-based or broadly focused campaign, you can specify the service lines and/or visit types on which we should base our list of suggested models. -OR-
• If your campaign’s objective is more granular/specific, you can specify the ICD-10 diagnosis, ICD-10 procedure, and/or MSDRG medical codes on which we should base our list of suggested models.

Once you arrive at a list of suggested models, you select the models to apply to your target audience, and using an interactive graph, define the risk levels (category codes) of the individuals to include.

Model Selector How-to.

Risk model suggestion logic

In order to determine the risk models to suggest, WebMD Ignite performed the following data analysis:

• Typical Volume. We aggregated a collection of historical visit data by medical code (ICD-10 diagnosis, ICD-10 procedure, and MSDRG); by service/sub-service line; and by visit type. This allows us to understand typical volume for a given medical code, service/sub-service line, and visit type.
• Model Mapping. We mapped medical codes, service lines, and visit types to appropriate consumer risk and patient risk models. This allows us to suggest an appropriate model per medical code, service/sub-service line, and visit type.

During the Risk Model Selector workflow, we compare the typical volume of all the risk models mapped to the medical codes OR the service lines/visit types you selected and offer risk model suggestions as follows:

• We suggest those models whose typical volume is historically associated with at least 5% of visits.
• Of the suggested models, we label a model as Best Fit when its typical volume is historically associated with at least 50% of visits.