The examination of physiological signals plays an important role in various fields, such as medicine, psychology and sports science. One method frequently used to collect physiological signals is photoplethysmography (PPG), which is used in particular to measure heart rate. However, the PPG signal is usually superimposed by disturbances. Therefore, the signals must be processed before they can be analyzed. Several toolboxes already exist for analyzing and processing physiological data. These are designed to enable users without in-depth knowledge of signal processing to easily perform accurate analyses. However, this raises the question of what influence the program packages and their default settings for preprocessing physiological signals can have on the results. To answer this question, the presented work first performs an analysis based on synthetic signals. These allow various use cases to be simulated and comparative signals to be generated to assess the effects of different analysis methods and parameters. This synthetic data is modeled on examples from stress training in virtual environments, since the investigation is carried out in this context. The context provides physiological parameters that can be used to quantify the effects. The work describes (1) which conditions should be met to be able to use synthetic PPG data for the selected context, (2) which dependencies exist between the selected physiological parameters and the used analysis methods and parameters and (3) which changes for these physiological parameters can be achieved by optimizing the analysis methods and parameters for selected example situations. For the generation of synthetic PPG data in the selected context – based on collected real PPG data – the combination, distribution, and strength of various artifact types over the synthetic PPG signal prove to be of crucial importance. When processing synthetic PPG data with the methods of a selected toolbox, even with low artifact strength, recognizable dependencies of the context-relevant physiological characteristic values on the analysis configuration used can be seen. Optimizing the analysis configuration for synthetic data results in significant advantages compared to the standard parameters of the selected toolbox for the same physiological characteristic values. The investigations into the transferability of the optimized solutions to similar and independently generated synthetic signals show that this transferability can depend on the similarity between the initial signal used for the optimization and the target signal to which the solution is applied. In combination with these results, a toolchain approach is presented in this work that uses synthetic substitute signals for real signals in conjunction with a synthetic artifact-free base signal. The goal of this approach is to optimize the selection of analysis methods and parameters on synthetic signals that are tailored to the individual and specific signal and artifact characteristics of the real signal to be examined. In tests with purely synthetic data, the optimized analysis configurations prove to be robust to a limited extent against approximation inaccuracies that may occur when the synthetic replacement signal does not exactly replicate the real signal. The presented toolchain approach can be used and further developed as a foundation for optimizing analysis configurations using synthetic data for preprocessing PPG signals and beyond. As complementary research directions, the introduction of additional methods, such as machine learning models, as well as advancing the development of synthetic data generation of the replacement signal with its associated artifact free base signal, can be suggested. In addition, evaluation procedures must be established that enable the added value of analysis configurations determined on synthetic data to be assessed when applied to real signals.     «

The examination of physiological signals plays an important role in various fields, such as medicine, psychology and sports science. One method frequently used to collect physiological signals is photoplethysmography (PPG), which is used in particular to measure heart rate. However, the PPG signal is usually superimposed by disturbances. Therefore, the signals must be processed before they can be analyzed. Several toolboxes already exist for analyzing and processing physiological data. These are...     »