• Login
• Admin

Tan, Alfian (2024) Modelling the variation and the reliability of a clinical process. PhD thesis, University of Nottingham.

PDF (Corrected Thesis) (Thesis - as examined) - Repository staff only until 10 June 2025. Subsequently available to Anyone - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (9MB)

Abstract

This research is conducted to propose a modelling approach to studying variations in a clinical procedure by also integrating a possible automation technique for identifying the variations. The modelling involves both graphical and simulation techniques, which accommodate a probabilistic analysis of the clinical procedure. This explores a new learning strategy on how somebody could experiment with the procedure before carrying on with a clinical experiment. To demonstrate this idea, this research studies the Newborn Life Support (NLS) procedure. The way this clinical procedure is analysed using this approach suggests a novel idea that contributes to the understanding of the interactions of elements in the NLS algorithm and of how far the approach of system modelling can be useful in this area.

The NLS procedure is an evidence-based protocol to resuscitate and stabilise a compromised baby. This protocol is usually performed in a team which consists of staff with different expertise and experiences. Variations in the performance of this procedure are influenced by both technical and non-technical aspects of the procedure. Hence, in this research, both of these aspects are investigated from in-field observation, literature, and expert interviews to result in a useful NLS model. Coloured Petri Nets (CPN) is used to first describe the logic of the NLS procedure in detail. A simulation model is then built following the CPN. This research yields two NLS models which involve the technical aspects of the procedure and the extension of the model that depicts the non-technical aspects of the procedure.

The development of the automated variations identification system is based on the combination of automated image segmentation and action recognition techniques. It aims to automate the analysis of NLS video recordings by automatically identifying the existence of an NLS step and its duration. It is beneficial to reduce the human efforts in the NLS performance evaluation and monitoring. The U-net Deep Learning structure for image segmentation and the learning strategy of using traditional machine learning models for the action classification step are used. The final variation recognition system resulting from this work still has a limited practical ability of action recognition. A significant additional dataset for every action category is definitely required.

Despite this limitation, this research is still concluded by showing how the automated variation recognition system and the NLS model can be integrated. The wet towel removal step is chosen to demonstrate this integration. The conceptual workflow of the information extraction and feeding process of this action between the recognition system and the NLS model is defined. It is realised by developing an integrated computerised system which is presented by a simple graphical user interface where the end user can determine their own NLS setting and observe the workflow of the system. Accelerating the computational process is essential to making this integrated process more beneficial in the future.

Item Type:	Thesis (University of Nottingham only) (PhD)
Supervisors:	Remenyte-Prescott, Rasa Egede, Joy
Keywords:	Healthcare Reliability; Healthcare Modelling; Newborn Life Support; Healthcare Simulation; Coloured Petri Nets; Image Segmentation; Action Recognition
Subjects:	R Medicine > R Medicine (General)
Faculties/Schools:	UK Campuses > Faculty of Engineering UK Campuses > Faculty of Engineering > Department of Mechanical, Materials and Manufacturing Engineering
Item ID:	79880
Depositing User:	tan, alfian
Date Deposited:	10 Dec 2024 04:40
Last Modified:	10 Dec 2024 04:40
URI:	https://eprints.nottingham.ac.uk/id/eprint/79880

Actions (Archive Staff Only)

Edit View