From the introduction:

Predictive models are data-driven, decision-support tools that estimate an individual’s future potential health care costs and/or opportunities for care management. Most commercially available PM tools classify individuals into future cost categories with a focus on high-cost cases. A few tools add a second component — commonly referred to as “impactability” — to identify patients who will potentially benefit from care management. Adapting PM tools to address the Medicaid population’s intense and complex array of needs — ranging from physical and behavioral health comorbidities to socioeconomic issues — is a critical consideration for states that are planning to use PM. This holds true for states buying off-the-shelf tools as well as those with the analytical capabilities to build and/or customize PM tools in-house.

Full report: Predictive Modeling: A Guide for State Medicaid Purchasers (PDF | 459.33 KB)

Center for Health Care Strategies, Inc. (2009). Predictive modeling: a guide for state Medicaid purchasers. Knutson, Dave; Bella, Melanie; and Llanos, Karen.