A drop in arterial pulse pressure below 65 mmHg can be life-threatening for a patient under general anesthesia during surgery, making rapid stabilization crucial. However, identifying the cause of this drop is not always immediately possible, as anesthesiologists often lack precise insights into the condition of the cardiovascular system. With the development of non-invasive blood pressure measurement methods, new techniques for analyzing arterial pulse waveforms have emerged, offering valuable insights into the patient's condition. This master's thesis aims to explore the potential of evaluating cardiovascular status based on waveform characteristics. Synthetic data, generated through a mathematical model of the cardiovascular system was used for this analysis. Although studies based on synthetic data have limitations, they can still play an important role in the development of analytical methods. Using principal component analysis, L1-regularized regression, and the XGBoost algorithm, we assessed how well waveform shapes could be used to estimate the parameters with which the waveforms were simulated, thereby describing the cardiovascular system's state. Finally, we also examined how effectively the system’s state could be described solely by observing the space of simulated waveforms and feature values.
|
