Blog

Deerwalk Plan Analytics Version 10.4 Uses Machine Learning to Predict Medication Adherence

Posted by Deerwalk Engineering Team on September 16, 2020

Deerwalk's latest release adds a new machine learning model and continues the phased rollout of our application redesign 

Release 10.4 blog image

Below are the key highlights from this month’s release of Deerwalk Plan Analytics:

  1. NEW MACHINE LEARNING MODEL: MEDICATION ADHERENCE
  2. NEW DATA ACCESSIBILITY CONTROLS
  3. NEW LOOK & FEEL (UI/UX UPDATES)

1. NEW MACHINE LEARNING MODEL: MEDICATION ADHERENCE 

We've released another machine learning model, the Medication Adherence Model, which helps you identify opportunities to increase medication adherence in your population(s). Using a number of factors, the model predicts the likelihood that a member will adhere to a newly prescribed medication over the next year (a medication is considered “new” if the member has never had a claim for a medication within that therapeutic class until the most recent month for which claims data is available).

You can pair the results from this model with Quality Metrics that are related to medication adherence to gain an even deeper understanding of a member’s overall risk. This information can also be very useful as you look to manage health and control costs for members taking medication to treat chronic disease. Having these insights will allow you to intervene early on, which is essential as initial adherence or non-adherence to a new medication is often indicative of a member’s future adherence patterns.

  • Adherence is determined by the proportion of days covered (PDC), which calculates the ratio of number of days the member is “covered” with medication (i.e., the number of days’ supply they’ve filled) in a period to the total number of days for which that medication has been prescribed in that same period. A member is considered to be “adherent” if the PDC is 80% or greater.
  • A probability score greater than 0.5 indicates a member is likely to adhere to a medication while a probability score less than 0.5 indicates a member is not likely to adhere to a medication.

 

2. NEW DATA ACCESSIBILITY CONTROLS

We've added a section that gives administrative users the flexibility to enable/disable data types for groups of users with different levels of access. A new settings panel called "Data Types" allows you to simply check a box to set data type permissions by each user tier. This gives organizations greater control over who has access to view and drill down into certain categories of information.

Data type access

3. NEW LOOK & FEEL (UI/UX UPDATES)

With the phased rollout of the usability and user interface redesign, we’ve made improvements to certain interface elements in several Plan Analytics modules, adjusting fonts for readability, and updating the appearance of pop-up windows and population reporting labels.

Med adherence popup

 

Schedule a demo to learn more about how to use Deerwalk Plan Analytics and the latest features.

REQUEST A DEMO

Subscribe to our blog or follow us on Facebook, Twitter or LinkedIn to join the conversation.

SUBSCRIBE TO OUR BLOG