Modifying sensitivity/specificity for sensors using positive and negative predictive power measures

Abstract

In the collection of data from sensors in the field, the uncertainty in the data may compromise the ability to accurately predict the state of a system. Herein the standard signal detection theory problem is examined when nonstationary effects may occur in the data from the sensors. The use of PPP (positive predictive power) and NPP (negative predictive power) adds a new viewpoint on how to modify sensitivity and specificity measures in decision making involving multiple sensors. This is especially true when stationary properties in received data may be violated.