Remote vital signs monitoring for early detection of deterioration after surgery
Mathilde van Rossum-Hermans is a PhD student in the department Cardiovascular and Respiratory Physiology. (Co)Supervisors are prof.dr.ir. H.J. Hermens and dr. Y. Wang from the faculty of Electrical Engineering, Mathematics and Computer Science, prof. C.J. Kalkman, UMC Utrecht and dr. E.A. Kouwenhoven, ZGT.
With the growing availability of wireless vital signs monitoring systems, opportunities for remote and continuous monitoring of patients within and outside the hospital arise. In surgical patients, remote vital sign monitoring may support improved recognition of postoperative complications and contribute to better patient outcomes and care efficiency. This thesis aimed to explore the desired design and expected clinical impact of remote vital signs monitoring strategies in surgical patients and to evaluate the performance of currently available remote sensing and decision-support techniques. Based on the results of our studies, we concluded that care professionals see opportunities to implement remote monitoring as an alternative or supplement to current in-hospital monitoring. Remote vital signs monitoring is expected to contribute to early detection of clinical deterioration in high-risk surgical patients in a ward or out-of-hospital setting and can allow early hospital dismissal, but only under certain circumstances and with suitable technology. Currently available wearable systems can monitor a selection of vital signs, but the measurement accuracy and reliability vary between systems and between vital parameters. Although imputation techniques may support further data analysis in case of missing data periods, we found that imputation errors vary strongly within and between imputation techniques and can affect clinical decision-making. Last, we showed that alarm systems that use adaptive thresholds or Early Warning Scores for the detection of abnormalities in continuous vital signs data may support early identification of complications in postoperative ward patients and perform better than classical single-parameter alarm systems, but also bring a serious risk of excessive false alarm rates. Careful selection and further improvement of sensing and data preprocessing techniques, decision-support methodology, and alarm settings are therefore warranted to ensure effective, reliable, and efficient patient monitoring.