Reliability models of chronic kidney disease

With the rise in quantifiable approaches to health care, lessons from reliability modeling provide new avenues for improving patient outcomes. Describing the development of conditions leading to organ system failure provides visceral motivation for quantifying reliability, or project outcomes in the face of uncertain disease progression. As prime examples, progressive kidney diseases leading to renal failure place increasing strain on finite health care resources. We propose a method of projecting future health states and cost by means of transient and cumulative numeric solutions of a homogeneous continuous time Markov chain. This publication constructs a model based on the clinical guidelines for kidney disease diagnosis, then populates the model with parameters from publically available data for the most general-case patient. Subsequently the model is solved for measures such as survival rate and expected cost incurred by a patient in a one year interval. Utility of the method is demonstrated in estimating future patient health and costs incurred at an individual level as well as how the results scale for practitioners and health care administrators. Results for the most general case population are presented, with recommendations for tailoring this method to specific patients. Finally we contrast this approach to similar studies, highlighting many of the common assumptions used in predictive models and in forecasting kidney disease prevalence to outline directions for future development of personalized models for prognosis.