Use this URL to cite or link to this record in EThOS:
Title: Hybrid physiological modelling and intelligent decision support for treatment of septic shock patients in cardiac intensive care unit
Author: King, Olufefni Kwasi
ISNI:       0000 0001 3599 8403
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2007
Availability of Full Text:
Access from EThOS:
In this information dominated era, computers have proved invaluable In the monitoring and the processing of signals, especially in sensitive environments such as Intensive Care Units and operating theatres. Despite the widespread use of systems engineering and control techniques in various industries, the benefits have usually extended to biomedical processes last, possibly due to concerns over safety. Where this can be established, applications tend to proliferate, as can be seen over the past few decades with the control of depth of anaesthesia, diabetes, respiration etc. However, only a few applications have been reported dealing with Septic Shock/Sepsis. Sepsis, the inflammatory state caused by infection, is a leading cause of death in Intensive Care Units worldwide. Patients in the United Kingdom are quoted a 2% through 20% risk of serious injury, stroke or death for each individual cardiac operation carried:out. The aim of the research work included in this thesis is to first describe the physiological behaviour of septic human patients via a hybrid model, by hybrid is meant a combination of different architectures, and second to exploit such a model by designing and implementing a decision support and control system to deliver the 'optimal' drug therapy which will reduce morbidity as well as mortality. The components of this research study will include the following overarching'themes: 1. A hybridphysiologically-based model is proposed/or the simulation ofaspects 0/ the cardiovascular system and its response to drugs and the onset 0/sepsis. The hybrid modelling strategy incorporates techniques from' cardiovascular modelling, microvascular exchange and pharmacokineticll?harmacodynamic (PKlPD) modelling to model cardiovascular blood flow, movement of fluid between the circulation and the interstitium and the response of the system to drugs respectively. Model suitability for describing real-life situations is tested by fitting model parameters to clinical data trends using a least-square technique. It is designed to be simple, yet physiologically relevant as this usually provides a greater insight into how a real life system works, as well as giving realistic results where real data are difficult to obtain. The completed model leads to a comprehensive platform for testing various control strategies in response to sepsis. 2. A multivariable self-organising fuzzy-logic controller (SOFLe) is developed to control drug infusionsfor post-CPB patients during sepsis. The SOFLC control scheme performance is improved with the addition of noise filtering and the functional range increased with the use of drug sensitivity estimation. The control scheme is extended to the multivariable case with the use of Relative Gain Array (RGA) decoupling technique. 3. The development of a comprehensive decision support and control structure for the management ofsimulated sepsis. Key to the development of this system is the acquisition of clinical expert knowledge, via a series of interviews and discussions. This gives rise to a decision tree structure which was represented as a hierarchical structure of fuzzy linguistic rules and implemented in a novel, comprehensive decision support and control structure. This is tested in simulation with closed-loop control of various states and evaluation of the decisions made.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available