Abstract: The objective of the proposed research is to develop robust tools to process noninvasively measured physiological signals using multi-modal wearable devices for hemodynamic monitoring. However, in ambulatory and noisy environments, such as those experienced during pre-hospital care in ambulances, physiological waveforms are severely degraded, resulting in unreliable monitoring systems. Therefore, processing techniques to reduce these artifacts and accurately extract features are integral to transition these systems to uncontrolled settings. These tools should also be operated in real-time to reduce latency. One signal modality that is especially affected by such artifacts is the seismocardiogram (SCG). This cardiomechanical signal, measured using an accelerometer on the chest, provides an aspect of cardiac function that other physiological signals fail to capture. However, the signal volatility of the SCG and its susceptibility to external vibrations necessitates robust processing tools for its use to be adopted outside of the lab setting. This work first investigates the effects of vehicular vibrations on SCG signals, designing a framework to separate the two. To achieve this separation, the periodicity of the SCG is harnessed for signal decomposition and rejection of non SCG-related components. Subsequently, we explore real-time SCG feature extraction techniques that could be used in these environments. The shared dynamics between SCG features are utilized to track these features using a Gaussian mixture model (GMM) framework. The proposed future work targets leveraging the shared information from other sensor modalities and improving real-time separation of external artifacts from SCG signals.