At RxAnte, we are often asked whether we use artificial intelligence (AI) as part of our adherence solution. The short answer to this question is ‘yes,’ but there is more nuance to this tool we find timely to explore; after all, AI is a popular concept—and it has become even more common with the rise of tools like Siri, Alexa, and autonomous vehicles.
Use of AI in Medication Adherence Solutions
In the context of medication adherence, RxAnte has pioneered the use of AI. Through our use of machine learning—a technique that falls under AI—we have delivered on a long-standing promise of anticipating medication non-adherence before it occurs. This is achieved through our use of predictive models, which are refined over time through machine learning. In fact, over the course of nearly a decade, our teams have experimented with a variety of predictive models to develop the most accurate method to anticipate medication use.
Based on our experience managing tens of millions of lives, however, we feel it is important to remember that the predictive analytics that we deliver as an outcome of our refined analysis is a tool. Despite the appeal of buzzwords like “AI” and “machine learning,” neither of these concepts are the end-goal. Our aim is to help our clients realize the greatest improvement possible in their adherence rates, and we believe in order to achieve this goal, it requires additional insights and tactics.
Incorporating a More Human Component
Improving medication adherence requires determining the most effective combination of selection criteria, member prioritization, and—critically—the best intervention for each member. Using predictive analytics to identify non-adherence does not fully address dynamic issues related to a health plan member’s receptivity to interventions or protect against member abrasion. This level of consideration requires frequent discussion between the internal team managing the health plan’s adherence program and RxAnte’s client services and data science teams.
Given this reality, our client programs have evolved from only predicting non-adherence to incorporating intervention receptivity, to now including simulation results as well. This approach is intended to support the most effective intervention. We surround these services with subject matter experts in data services and analytics, and supply a dedicated client services team that works closely with each client to continue to refine the program toward achieving optimal results. The analytics we deploy are ultimately tailored to clients’ specific data sets, and further refined to account for the most effective interventions.
Evidence illustrates the effectiveness of this approach: year over year, our clients exceed industry performance in adherence improvement. In fact, in 2018, RxAnte client improvements in diabetes medication adherence nearly doubled the broader industry’s percent improvement for comparable 3-Star plans. So while machine learning continues to inform predictive models, their outputs are one piece of a broader solution that has provided our clients with reliable improvements to their adherence programs.
Interested in learning more about how RxAnte’s approach to medication adherence can be part of your solution? Contact us.