Real-time, non-additionally invasive metrics of cardiovascular performance in critical care: a model-based framework
Thesis DisciplineMechanical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Cardiovascular disease and dysfunction (CVD) are leading causes of Intensive Care Unit (ICU) admission, costs, and mortality worldwide. Aging populations demand increasingly personalised and optimised cardiovascular care in the ICU to meet demand for more care at lower costs. However, inadequate or incorrect diagnosis of cardiac disturbances resulting in increased length of stay, cost and mortality is an ongoing issue.
Cardiac management in the ICU is informed by measurements taken from a variety of instruments, such as catheters placed near the heart. However, despite the rich information available from such instruments, their use is not necessarily associated with improved clinical outcomes. This discrepancy may be due to the clinical simplification of instrument outputs into lumped metrics that can be easily and rapidly parsed, but fail to preserve much of the initially gathered, patient-specific information. Thus, patient-specific information is lost and care is titrated based on population driven metrics, reducing the potential quality of care. Hence, improvement in the extraction of patient-specific cardiac information from these instruments has the potential to yield significant patient-centred, social, and economic value from clinically available data that has potentially been under-utilised to date.
This thesis develops a patient-specific modelling framework for the non-additionally invasive clinical estimation of a number of key, interlinked, and patient-specific cardiovascular metrics, including the end-systolic pressure-volume relation (ESPVR), pressure-volume (P-V) loop, unstressed or ‘dead space’ ventricular volume (Vd), time-varying elastance (TVE), and end-systolic elastance (Ees). These metrics are already accepted parts of clinically standard frameworks of understanding the underlying physiological behaviour of the circulatory system. However, they are not currently available in the ICU, forcing clinicians to make links between simple, observed parameters and these metrics using their own mental models of understanding, experience, and intuition. Hence the provision of these particular patient-specific metrics from clinically available measurements in the ICU has the potential to provide significant benefits in clinical decision making, and personalisation of care, at little additional cost by leveraging models and computation.
This thesis further explores directly extracting information about the condition of the systemic circulation from various commonly measured pressure waveforms. In particular, relationships between the aortic, femoral and central venous pressures are explored through the medium of both transfer functions and a simple waveform energetics framework. The goal of this additional research is to provide further simple, intuitive and clinically sensitive metrics of systemic circulation condition from commonly available clinical data.
Each area of study is validated across two experimental animal trial data sets. These data sets encompass several popular clinical interventions in cardiac and circulatory management, such as fluid resuscitation, the administration of inotropic drugs, and recruitment manoeuvres. They also include an animal model of septic shock, which is a leading cause of acute circulatory and/or cardiac failure. In each case, a wide variety of information was invasively measured, allowing direct validation of model generated outputs against direct, invasively measured equivalents. Additionally, the interventions and conditions selected were designed to provide a diverse and challenging range of cardiovascular states, ensuring an overall robust and rigorous validation of each metric, model and method developed.
Overall, this thesis provides a coherent, interlinked framework for the non-additionally invasive model-based estimation of patient-specific and time-specific cardiovascular metrics central to titrating clinical care. The provision of these metrics in the ICU has the potential to optimise patient monitoring, aid real-time clinical decision making at the patient bedside, and provide personalised care, closing the gap between the conceptual models of understanding a clinician relies on, and the surface level measured and summarised metrics clinically available. In essence, using available data, models and computation to create a much clearer physiological and patient-specific picture for clinicians to use in care and management.