Cardiovascular Computing in the Intensive Care Unit

Abstract

Clinical practice in the intensive care unit (ICU) faces a number of challenges, including accurate and early detection of pathological processes, and the related decision-making often relies on haemodynamic monitoring. In this chapter, applications are presented of computerised approaches for analysing data obtained from haemodynamic monitoring in the ICU. Haemodynamic monitoring is primarily concerned with assessing the performance of the cardiovascular system and conventionally relies on blood pressure measurements and echocardiography, for estimating cardiac output and other physiological variables. In addition to haemodynamic monitoring, less common techniques in the ICU (applanation tonometry, carotid and venous ultrasound), can also be used in cardiovascular computing applications. Such applications span a wide range of clinically relevant issues, including organisation and archiving of data into structured databases, data analytics, decision making and prediction, as well as estimation of arterial stiffness. Large, comprehensive, publicly available databases facilitate benchmarking of machine learning algorithms using real-world data. Such algorithms can in turn contribute to improving sepsis prediction in the ICU through (a) the identification of new features useful for prediction and (b) the processing of large data amounts, so as to consider combined contributions of individual features. The evidence produced so far indicates that cardiovascular data archiving and analysis using advanced computing methodologies is promising for addressing crucial issues in the ICU, and highlights the role that clinical data analysis will increasingly play in both knowledge generation and medical practice.