Modelling and control of hemodialysis systems for better treatment management of Chronic Kidney Disease patients

• Author / Creator

  Gnanasekar, Aristarchus

• Kidneys are essential organs located on either side of the vertebral column which perform several essential bodily functions. When there is a gradual, permanent loss of basic kidney functions, a person is said to have Chronic Kidney Disease (CKD). CKD has been identified to be a global public health issue affecting millions of people every year. CKD can progress to an end-stage, and the patient's life would be at stake without artificial filtering (dialysis) or a kidney transplant. Hemodialysis is a life sustaining treatment for End Stage Renal Disease (ESRD) patients. Though being the most frequently used treatment modality, there are numerous clinical complications, while the most common include Intradialytic Hypotension (IDH) and Dialysis Disequilibrium Syndrome (DDS), arising during the quick extra-corporeal depuration of blood in an external device called ‘hemodialyzer’, which is sometimes referred to as ‘artificial kidney’. This thesis starts off with a literature review in chapter 1 followed by a technical preliminaries review in chapter 2, to help the readers understand the research background and the problem better.

  The perturbations caused by hemodialysis in a patient’s body are complex, though the underlying phenomenon is a simple bidirectional mass transfer. The use of a mathematical model can enable a quantitative analysis of perturbations (cardiovascular response, fluid and solute kinetics) induced by hemodialysis taking place within the patient’s body in different hemodialysis treatment settings and can help in understanding the intricate physiological mechanisms. In chapter 3, the mathematical models selected for representing each of the hemodialysis subsystems are presented along with some derivations, assumptions, control relevant modifications along with some simulations representing the hemodynamics of different classes of CKD patients.

  Each patient behaves differently to hemodialysis and the challenge is to achieve meaningful predictions for each individual patient. Chapter 4 talks about the design of a simultaneous state and parameter estimation algorithm, specifically intended to identify individualized virtual patient simulators, based on synthetic clinical data, which could aid prediction of important state variables like Mean Arterial Pressure (MAP), Heart Period (HP), etc.,. For consistent estimation, the observability of the system has to be ensured and the nonlinear system observability test is not as straightforward like it would be for linear models. In our proposed approach, a sensitivity-based local observability test shall be conducted. The sensitivity equations should be solved in parallel with the original model equations to obtain the sensitivity matrix. Then a singular value decomposition is done to obtain the observability signature graph. A clear drop in the graph indicates a lack of observability. If such clear drops are encountered the user has to identify a subset of observable variables from the total variable set for estimation. For this purpose, a sequential orthogonalization algorithm was applied, to forward select the non-correlated variables one at a time until the terminating conditions are met, starting from the most sensitive and least correlated variable. The returned subset would be the decision variables during the simultaneous state and parameter estimation routine.

  Traditionally, the hemodialysis treatments are done in open-loop fashion where the treatments are stopped when clinical complications occur and started again after the patient returns to normalcy. The model thus identified from chapter 4 could be used to design ‘individualized optimal treatments’ using advanced model based controllers, like a Batch Zone Model Predictive Controller (BZMPC) with a built in nonlinear state estimator, with feedback implementation while taking the treatment objectives and safety constraints into account as discussed in chapter 5, thus paving the way for continuous optimal safer treatments. Finally, the future research directions are narrated in chapter 6.

• Subjects / Keywords

  • hemodialysis
  • modelling
  • control
  • treatment
  • chronic
  • kidney
  • disease
  • management
  • better

• Graduation date

  Fall 2021

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-crh5-ra34

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.