Bayesian Decision Analysis for Diagnostic Procedures

• Activity Number: 107
• Type: Invited
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Committee on Applied Statisticians
• Author(s): Ying Lu*+ and John Kornak
• Companies: Stanford University and University of California, San Francisco
• Keywords: Baysian Decision Making ; Osteoporosis ; Bone Mineral Density (BMD) ; Utility ; Cost Effectiveness

Abstract

New diagnostic procedures are continually being developed with the competing goals of increasing accuracy and reducing costs. Decisions are needed for introducing a new procedure or to continue with an existing one when both goals are in conflict. The decision is complicated because diagnoses are often based on thresholding one or more continuous variables. Therefore the decision process should account for uncertainty in the threshold value. To address these issues, a Bayesian decision theoretic approach is developed that defines the optimal decision based on the maximization of expected utility (incorporating accuracy and costs) with respect to diagnostic procedure and threshold level. The Bayesian decision approach is illustrated via an application osteoporosis, which demonstrated to a powerful strategy for determining optimal diagnostic procedures in the presence of uncertainty.