S2: Acquisition of physiological parameters at the prosthesis interface

• Applicant: Professor Dr.-Ing. Hagen Malberg
• Subject Area: Biomedical Systems Technology
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 550772260

The prosthesis-skin interface is subject to mechanical and thermal stresses and can cause problems such as pain, limb atrophy and tissue damage. Given these complex dynamics, an in-depth biomechanical understanding of the internal interface environment is essential, specifically with regard to the interfascial stress profile. This is critical to minimize the risk of injury and tissue damage and to increase prosthesis acceptance rates. A monitoring system should contribute to the early detection of injury risks and allow automatic adjustment of the prosthesis stem geometry. With the current state of the art, it is not possible to quantify and localize day-dependent volume fluctuations in the tissue. The aim of FOR 5493 is to optimize prosthetic fitting taking into account intraindividual tissue variability by means of sensor-based interface analysis and adaptation of the prosthetic socket. The subproject focuses on a multimodal sensor system that captures the complexity of the interface interaction by objectively measuring different biosignals. The sensor system is to be integrated directly into the interface as an inlay. In order not to compromise comfort in the process, the most promising approach appears to be to build the sensor array entirely from an elastic polymer. The realization of a sensor array with the lowest possible degree of additivation (≤ 5 mass %), in order to maintain elastic polymer properties, as well as a miniaturization of the electrodes and conductive tracks is the research object of this subproject. Everyday load scenarios are simulated on a test bench in order to experimentally investigate the robustness of the sensor technology. Sensor signals are characterized with respect to their signal morphology and heterogeneity of the signal distribution. Further, exploratory investigations using feature engineering will search for innovative features to comprehensively evaluate the state of the residual limb. The key to localizing volume fluctuations is the processing of the extracted signals. These are first evaluated separately in neural networks, higher-level features are extracted via network separation, and these are trained comparatively in a multi-branch approach. Transfer learning will be used to optimize the model for project-specific load scenarios. At the end of the project, the localization of the volume fluctuations will be achieved, which will be used as transfer parameters by the FOR research partners. As a long-term goal, the subproject of FOR 5493 is to create both an innovative sensor setup and new methods of signal evaluation, which can be further investigated in future experiments on patients in dynamics.

• DFG Programme: Research Grants