Practice of Management Science—Patient States and the Technology Matrix

This paper describes the procedure used to estimate transition probabilities from a combination of expert judgment and empirical evidence using a Bayesian model. The matrix of transition probabilities (called the Technology Matrix) represents the Markovian process by which changes in states (Patient States) occur in the target population subsequent to receiving a specified treatment (Service Package). The overall model in which the Technology Matrix is embedded is described briefly. The data collection process and limitations on data are outlined and the Bayesian procedure is developed. Results obtained from initial observations are used to illustrate the procedure.

Duke Scholars

Author Richard M. Burton Fuqua School of Business