Prioritizing high-need, high-cost Medicaid populations by risk and level of need can help states and health plans identify “high-opportunity” beneficiaries and target quality improvement resources appropriately. While predictive modeling tools have historically been used to predict costs for rate-setting purposes, states can also use these tools to identify “high-opportunity” candidates for care management and target public resources more effectively.
To help state Medicaid agencies use predictive modeling tools to identify and prioritize candidates for care management, the Center for Health Care Strategies (CHCS) partnered with David Knutson from the University of Minnesota to develop Predictive Modeling: A Guide for State Medicaid Purchasers. The guide outlines key considerations for states to address prior to purchasing or building a predictive modeling tool.
States interested in implementing predictive modeling can use this guide to:
- Understand which features of a predictive model are critical as well as how to enhance information that is derived from predictive models for Medicaid populations;
- Address planning questions to guide the implementation of predictive modeling; and
- Outline key considerations for choosing a predictive modeling tool to identify candidates for care management.
