Announcement: SOA releases March 2022 Exam P passing candidate numbers.

E&R Sessions at 2008 Health Spring Meeting

E&R Sessions at 2008 Health Spring Meeting

By Steve Craighead

The 2008 Spring Heath Conference will be held in at the Hyatt Regency Century Plaza in Los Angeles. Described below are the scheduled sessions that are either sponsored or co–sponsored by the E&R Section.

  1. The Stochastic and Artificial Intelligence Models in Health Insurance Continental Breakfast will be held on Thursday, May 29 from 6:30 a.m. to 7:50 a.m. Its session number is 29 and its value ladder stages are Task/Technical and Process. This session is open to all meeting attendees and is free of charge. Here is a description of the session:

    • Many advanced statistical techniques are available for use in modeling health benefits. These methods are well known to most statisticians, but not most actuaries. A few examples are Bayesian, Markov chain, generalized additive models, generalized linear models and non–parametric models. Also, there are some non–statistical artificial intelligence techniques such as neural nets. The presenters will show how these methods can improve health insurance analyses in applications such as projecting health insurance claims, reserving and pricing health benefits. At the conclusion of this session, you will have an understanding of what these procedures can do to aid in getting more accurate results. You will also gain a basic understanding of how to perform the techniques and will be provided resources to gain a more thorough understanding.
  3. The "Predictive Models: Paralyzing Puzzle or Process Panacea?" session will be held on Thursday, May 29 from 10:00 a.m. to 11:30 a.m. This session is jointly sponsored with the Health Section Council. The moderator is Vincent Kane, FSA, MAAA. The session number is 45 and its value ladder stage is Process. Here is a description of the session:

    • Predictive models have been implemented across health care organizations to measure provider efficiency, develop risk–adjusted payments, project costs for pricing and underwriting, and to profile high cost cases for medical management. However, different applications require different types of predictive models that were developed to answer specific questions of interest.
    • These models vary according to underlying theory, the calibration data sample, the predicted outcome, data inputs feeding the model and the method of estimation. Such aspects must be considered before a model is ever implemented so that its results are accepted by stakeholders. Correct model selection leads to results which facilitate, rather than paralyze, decision–making.
    • Participate in this session to gain a deeper insight into the appropriateness of specific predictive models for tailored applications, recognizing that "turnkey" or "one–size–fits–all" models often do not generate the results one would expect. In addition, you will be equipped to identify the drivers of variability in health care resource use (e.g., chronic– versus–acute care) and how this individual heterogeneity affects model performance measures. Other topics covered in the session include the evolution of applications and acceptance of predictive models going forward.

We hope to see you there!