Article
Estimating Member-Level Treatment Effects with Uplift Modeling
By Randy Beams and Tanner Boyle
03 December 2024
In the dynamic landscape of long-term care, insurers are moving to adopt value-based care principles and continually seeking innovative ways to proactively manage their blocks. In an article published with the Society of Actuaries, we explore powerful tools that can enhance traditional effectiveness studies and address some of their existing weaknesses. One approach involves uplift modeling, where one can directly estimate the effect of a treatment on each member individually. In the article, we discuss the following topics:
- Experimental design considerations
- Creating a synthetic control group
- Traditional effectiveness methods
- What is uplift modeling?
- Data-driven decision-making
- “Long-er” term care