TY - THES ID - 137258741 TI - Development of an algorithm for the monitoring of patients implanted with a mechanical circulatory support system AU - Kloudová, Aneta AU - Geris, Liesbet AU - Fresiello, Libera AU - KU Leuven. Faculteit Ingenieurswetenschappen. Opleiding Master of Biomedical Engineering (Leuven) PY - 2020 PB - Leuven KU Leuven. Faculteit Ingenieurswetenschappen DB - UniCat UR - https://www.unicat.be/uniCat?func=search&query=sysid:137258741 AB - Left ventricular assist devices (LVADs) are increasingly used with patients suffering from end-stage heart failure. Despite many advantages and significant technological improvement many complications are still associated with the LVADs post implantation. Therefore, several monitoring and control algorithms have been introduced for the detection of adverse events. The general aim of this thesis was the development of an algorithm for the monitoring of patients implanted with LVAD. The LVAD was connected to a cardiovascular simulator and different patient and pump conditions were simulated. The reproduced patient conditions included cardiac arrhythmias, change in left ventricular (LV) contractility, increased LV stiffness, change in afterload, and change in blood volume status. The simulated pump condition was a pump suction. Additionally, experiments with different pump speeds were realized to test the influence of the pump speed on the pump signal features. The developed monitoring software can correctly and noninvasively detect pump signal features indicative of changes in the patient hemodynamics and pump suction. For each simulated patient condition an analysis was done to evaluate which features change the most and how to identify a patient condition unequivocally. The only simulated patient condition which can be detected with high certainty based on the calculated signal features is the cardiac arrhythmia. The remaining simulated patient conditions may be detected by using a set of features with emphasis on their magnitude and clinical context. There was a distinction between the signal features detected during a normal pump flow and state of pump suction. The features utilized for the detection of the patient’s hemodynamics were influenced by different pump speeds. In conclusion, the thesis accomplished to prove the complexity of the pump-patient interaction and demonstrate that the patients hemodynamics significantly influences several features of the pump signal simultaneously. ER -