Efficient Value of Information Calculation Using a Nonparametric Regression Approach: An Applied Perspective

Background: Value-of-information (VOI) analysis provides an analytical framework to assess whether obtaining additional evidence is worthwhile to reduce decision uncertainty. The reporting of VOI measures, particularly the expected value of perfect parameter information (EVPPI) and the expected value of sample information (EVSI), is limited because of the computational burden associated with typical two-level Monte-Carlo–based solution. Recently, a nonparametric regression approach was proposed that allows the estimation of multiparameter EVPPI and EVSI directly from a probabilistic sensitivity analysis sample.

Objectives: To demonstrate the value of the nonparametric regression approach in calculating VOI measures in real-world cases and to compare its performance with the standard approach of the Monte-Carlo simulation.

Methods: We used the regression approach to calculate EVPPI and EVSI in two models, and compared the results with the estimates obtained via the standard Monte-Carlo simulation.

Results: The VOI values from the two approaches were very close; computation using the regression method, however, was faster.

Conclusion: The nonparametric regression approach provides an efficient and easy-to-implement alternative for EVPPI and EVSI calculation in economic models.