There is a growing body of literature that recognizes the importance of Skin Conductance (SC) for assessing changes in emotional states, such as engagement to learning tasks, and its importance to estimate possible drawbacks affecting overall performance. To date, most of the commonly used methods for SC signal analysis, i.e. detecting its phasic and tonic components and thus extracting informative features, are either too simple and unreliable or too complex and thus inaccessible and inflexible, as well as unable to perform online analyses. The current work proposes a simplified but clear and effective algorithm based on a Machine State to search for expected behaviors in the well-defined morphology of the signal. Eleven (11) features were correctly extracted from 79 healthy subjects during an experimental setup for immersive virtual rehabilitation (balance study case). The method was also successfully applied as a tool to identify significant changes in the subjective psychophysiological response to different experimental conditions. These results point toward a potential role in virtual rehabilitation applications by getting real-time feedback in human-in-the-loop approaches.