Fuzzy rule-based inference in system dynamics formulations

In this research, we broaden the scope of system dynamics formulations by building on a previously proposed approach to bridge fuzzy logic with dynamic modeling. Our methodology illustrates how to formulate fuzzy dynamic variables in a meaningful way. We highlight several modeling challenges, including the selection of a fuzzification and defuzzification method, their implementation in a system dynamics formulations and the validation of the results. We use a physician prescription decision-making model substructure as an example, and apply the fuzzy rule-based inference system to determine how a patient is categorized as "low-risk," "average-risk" or "high-risk." We emphasize various interpretation challenges and suggest careful selection of the fuzzy operators and defuzzification method, to ensure that the defuzzified values behave reasonably in a dynamic context. Copyright (c) 2020 System Dynamics Society