Modelling of expert nurses' pressure sore risk assessment skills as an expert system for in-service training

Hyslop, Alan (1988) Modelling of expert nurses' pressure sore risk assessment skills as an expert system for in-service training. PhD thesis, University of Glasgow.

Abstract

In the nursing literature to date there have been no reported applications of `cognitive simulation' nor of intelligent Computer Assisted Learning. In Chapter 1 of this thesis a critical review of existing nurse education by computer is used to establish a framework within which to explore the possibility of simulation of thinking processes of nurses on computer. One conclusion from this review which is offered concerns the importance of firstly undertaking reliable study of nursing cognition. The crucial issue is that an understanding must be gained of how expert nurses mentally represent their patients in order that a valid model might be constructed on computer. The construction of a valid computer based cognitive model proves to be an undertaking which occupies the remainder of this thesis. The approach has been to gradually raise the specificity of analysis of the knowledge base of expert and proficient nurses while seeking concurrently to evaluate validity of the findings. Reported in Chapter 2, therefore, are the several experimental stages of a knowledge acquisition project which begins the process of constructing this knowledge base. Discussed firstly is the choice of the skill domain to be studied - pressure sore risk assessment. Subsequently, the method of eliciting from nurses top-level and micro-level descriptors of patients is set out. This account of knowledge acquisition ends with scrutiny of the performance of nurse subjects who performed a comprehensive simulated patient assessment task in order that two groups might be established - one Expert and one Proficient with respect to the nursing task. In Chapter 3, an extensive analysis of the data provided by the simulated assessment experiment is undertaken. This analysis, as the most central phase of the project, proceeds by degrees. Hence, the aim is to `explain' progressively more of the measured cognitive behaviour of the Expert nurses while incorporating the most powerful explanations into a developing cognitive model. More specifically, explanations are sought of the role of `higher' cognition, of whether attribute importance is a feature of cognition, of the point at which a decision can be made, and of the process of deciding between competing patient judgements. Interesting findings included several reliable differences which were found to exist between the cognition of subjects deemed to be proficient and those taken as expert. In the final part of this thesis, Chapter 4, a more formal evaluation of the computer based cognitive model which was constructed and predictions made by it was undertaken. The first phase involved analysis in terms of process and product of decision making of the cognitive model in comparison to two alternative models; one derived from Discriminant Function Analysis and the other from Automated Rule Induction. The cognitive model was found to most closely approximate to the process of decision making of the human subjects and also to perform most accurately with a test set of unseen patients. The second phase reports some experimental support for the prediction made by the model that nurses represent their patients around action-related `care concepts' rather than in terms of diagnostic categories based on superficial features. The thesis concludes by offering some general conclusions and recommendations for further research.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science; R Medicine > RT Nursing
• Colleges/Schools: College of Science and Engineering
• Supervisor's Name: Jones, Dr. B.T. and Logan, Miss W.
• Date of Award: 1988
• Depositing User: Mrs Marie Cairney
• Unique ID: glathesis:1988-2932
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 17 Oct 2011
• Last Modified: 05 Feb 2014 15:47
• URI: https://theses.gla.ac.uk/id/eprint/2932