A Physiologically-Motivated Model of Creatinine and Fluid Volume Dynamics in Acute Kidney InjuryRichards, Evan (2018) A Physiologically-Motivated Model of Creatinine and Fluid Volume Dynamics in Acute Kidney Injury. Master's Thesis, University of Pittsburgh. (Unpublished)
AbstractBlood serum creatinine concentration (SCr) is used as a surrogate for kidney function. In an intensive care setting, SCr is used to estimate the extent of Acute Kidney Injury (AKI). AKI occurs in 42% of all intensive care unit (ICU) admissions, and it can lead to devastating impacts on the body including the development of interstitial and pulmonary edema, toxin accumulation, and excess mortality. Previous research shows the benefits of utilizing an absolute scale for measuring SCr and the necessity to consider the impact of systemic volume changes. The present work develops a biologically-motivated, low-order model of volume and creatinine dynamics that further progresses the understanding of an SCr measurement. Fluid volume is modeled into three interacting spatial compartments representing blood, interstitial volume, and intracellular volume. The blood compartment is further divided into plasma and liquid contained within the hematocrit (red blood cells). The four compartment creatinine model uses a similar structure to the fluid volume model, but combines the intracellular and interstitial volumes together and re-distributes these volumes between the muscular and non-muscular regions. The rate of creatinine generation is known to decrease during AKI. This understanding motivates the need for a model able to capture the influence of changes in rate of creatinine generation on the creatinine concentrations across the included compartments. A trajectory of creatinine generation rates is included in this work. Simulated studies of dehydration and fluid overload across six days demonstrates the ability of this model to capture kidney function changes in scenarios where absolute SCr measurement would not recognize AKI as promptly. The calculated differences are shown in hypothetical, clinical scenarios by varying levels of kidney function at differing levels of chronic kidney disease - none, stage 2, stage 3, and stage 4. To examine the applicability of this model to a clinical setting, its performance is tested by studying its aptitude to fit data collected from ten different patients at the University of Pittsburgh Medical Center (UPMC) and portray kidney function. Patient data fits accompanied by simulated studies conclude the importance of integrating human physiology into a trivial, low-order model that considers critical components of volume and creatinine dynamics. Share
Details
MetricsMonthly Views for the past 3 yearsPlum AnalyticsActions (login required)
|