The thesis discusses two methods for determining the direction of coupling of systems from their measurements. The first method is based on information theory and the second one on time-frequency transformation. The goal of the thesis is to test their usefulness for determining the direction of coupling of systems in different dynamic regimes and for solving the inverse problem of inferring the model parameters of coupling. We test them on the model of two coupled Duffing oscillators in a chaotic and a regular dynamic regime. For the systems in regular regime, the problem of determining the direction of coupling turns out to be simple, and it is also solvable in a chaotic regime. The inverse problem is solvable, except for the coupling of two chaotic systems. We also introduce a statistical test using surrogate data to confirm the existence of coupling. Finally, we discuss the application of the methods in medicine as a diagnostic tool.