Pin Me

Predictive Modeling in Health Care: Benefits and Issues of Using Modeling Tools

written by: GiangNguyen • edited by: Anurag Ghosh • updated: 11/10/2009

Predictive modeling has emerged as an important tool used for management of health care. In future, predictive modeling will become part of the electronic health care work-flow and help health care providers to maximize health benefits for individuals and populations.

  • slide 1 of 5

    Costs on the Rise

    With the costs of health care exploding in the last twenty years, predictive modeling is one of many tools that policymakers, insurance companies and health organizations turn to for help. One can use predictive modeling to estimate disease risk and to evaluate the effectiveness of a health care intervention. The models can be used to predict health care costs, utilization (such as hospitalization rate, ER/ED, medications etc.).

  • slide 2 of 5

    How Predictive Modeling Helps

    A great opportunity for predictive modeling is its ability to address challenges that various care management / disease management (CM/DM) programs are facing. It is believed that predictive modeling is critical for getting more value from these programs. It is now understood that risk and costs of health care are concentrated in a small proportion of the population. It is estimated that 50% of the population accounts for 80% of medical costs. Predictive modeling can be used to identify people with high medical needs who would likely benefit from care management interventions.

    Health care professionals have become more aware of the power of predictive modeling. Predictive Modeling News, the only newsletter dedicated to industry professionals involved with health care predictive modeling was introduced in January 2008.

  • slide 3 of 5

    Types of Modeling Tools

    Most predictive modeling tools can be classified into “rule set" based models and “data mining" based models. With the introduction of electronic health records and other health information technologies into standard health care practice, massive datasets will be generated in the next decades and can be used to build and to verify models. The models can be built based on information from diagnosis, pharmacy and prior-use data found in health insurance claims.

  • slide 4 of 5


    The main issues with current predictive modeling tools are lack of transparency and standardization. Most tools are used as black-boxes with limited understanding of the basic assumptions underlying the models. Unlike engineering simulation tools, predictive modeling software in health care are far from standardized. Furthermore, most models are built with a one-size-fits-all approach, while in reality, each organization has a unique business need.

  • slide 5 of 5


    Predicting Mortality and Healthcare Utilization with a Single Question, KB DeSalvo, VS Fan, MB McDonell, SD Fihn - Health Services Research, 2005 - Blackwell Synergy

    Principles of Good Practice for Decision Analytic Modeling in Health-Care Evaluation, MC Weinstein, BO'Brien, J Hornberger, J Jackson, M … - Value in Health, 2003 - Blackwell Synergy

    Modeling for Health Care and Other Policy Decisions: Uses, Roles, and Validity, MC Weinstein, EL Toy, EA Sandberg, PJ Neumann, JS … - Value in Health, 2001 - Blackwell Synergy