An Adaptive Identification Test Monitoring Procedure for Nonlinear Behavioral Interventions

Abstract Different studies have established correlation between physical inactivity and the incidence of chronic diseases. Prior investigations have been developed around the topic of mobile physical activity interventions relying on Multiple Input Multiple Output (MIMO) dynamical models of Social Cognitive Theory (SCT) that have been obtained through control engineering and system identification approaches. Identification Test Monitoring (ITM) is a technique that yields to the estimation of an adequate model with the shortest possible duration of the experiment. In this context, Local Polynomial Method (LPM) has been applied to estimate the Frequency Response Function (FRF) and the power spectrum of the disturbing noise for linear models. However, the experimental setup of physical interventions considers a decision block that is nonlinear. This paper describes the redesign of an ITM procedure for nonlinear behavioral interventions, through new uncertainty computations and stopping criterion analysis.