THE IMPACT OF EXPERIMENTAL DESIGNS & SYSTEM SLOPPINESS ON THE PERSONALISATION PROCESS: A CARDIOVASCULAR PERSPECTIVE

Abstract

To employ a reduced-order cardiovascular model as a digital twin for personalised medicine, it is essential to understand how uncertainties in the models input parameters affect its outputs. The aim is to identify a set of input parameters that can serve as clinical biomarkers, providing insight into a patients physiological state. Given the challenge of finding useful clinical data, careful consideration must be given to the experimental design used to acquire patient-specific input parameters. In this paper, we conduct the first quantification of a cardiovascular systems sloppiness to elucidate the structure of the input parameter space. By utilising Sobol indices and examining various synthetic cardiovascular measures with increasing invasiveness, we uncover how the personalisation process and the cardiovascular systems sloppiness are contingent upon the chosen experimental design. Our findings reveal that continuous clinical measures induce system sloppiness and increase the number of personalisable biomarkers, whereas discrete clinical measurements produce a non-sloppy system with a reduced number of biomarkers. This study underscores the necessity for careful consideration of available clinical data as differing measurement sets can significantly impact model personalisation. Author SummaryIn personalised medicine, computational models that replicate physical systems -- are becoming vital tools for understanding and predicting individual health. Our study explores cardiovascular models, which simulate heart and circulatory functions from which clinical metrics may be derived. These models aim to provide personalised insights into heart health and treatment planning. A key challenge in building these models is addressing "sloppiness," a property which provides vital insight into the response surface structure for which one calibrates a model searching for a global minimum point, a position in parameter space which best represents a patients cardiovascular health. In order to personalise a model different types of clinical metrics must be available for a model response to be compared to. We examined how different types of clinical data -- ranging from simple discrete blood pressure readings to detailed invasive continuous waveform data -- impact model sloppiness and the number of personalisable biomarkers. Our results show that continuous measurements increase the number of personalisable biomarkers but make the personalisation process more complex through increased sloppiness. In contrast, simpler discrete measurements reduce model sloppiness simplifying the personalisation process but yield fewer personalisable biomarkers. By analysing the impact of experimental designs on the personalisation process, our work offers practical insights into improving the reliability of cardiovascular digital twins, supporting their adoption in personalised medicine.