Use this URL to cite or link to this record in EThOS:
Title: An exploration of the adaptive neuro-fuzzy inference system (ANFIS) in modelling survival
Author: Hamdan, Hazlina
Awarding Body: University of Nottingham
Current Institution: University of Nottingham
Date of Award: 2013
Availability of Full Text:
Full text unavailable from EThOS. Restricted access.
Please contact the current institution’s library for further details.
Medical prognosis is the prediction of the future course and outcome of a disease and an indication of the likelihood of recovery from that disease. Prognosis is important because it is used to guide the type and intensity of the medication administered to patients. Patients are usually concerned with how long they will survive after diagnosis. Survival analysis describes the analysis of data that corresponds to the time from when an individual enters a study until the occurrence of some particular event or end-point. It is concerned with the comparison of survival curves for different combinations of risk factors. Analytical methods that are transparent for the clinician's understanding and explain individual inferences need to be considered when dealing with medical data. This thesis describes a methodology for modelling survival by utilising the application of the Adaptive Neuro-Fuzzy Inference System (ANFIS). A hybrid intelligent system which combines the fuzzy logic qualitative approach and adaptive neural network capabilities towards better performance. The ANFIS approach was applied in modelling survival of breast cancer based on patient groups derived from the Nottingham Prognostic Index (NPI). A comparison of the proposed method with the existing methods in the capability to predict the survival rate is presented. The use of a fuzzy inference system (FIS) in modelling survival is expected to offer the capability to deliver the process of turning data into knowledge that can be understood by people. The design of rules can be performed either by human experts or using appropriate approaches to build high quality PIS to represent the knowledge. In this thesis, represent an automatic generation of membership functions and rules from the data. Further, corresponding subsequent adjustments have been made to the model to give towards more satisfactory performance. The final premise and consequent parameters obtained are then used to predict the survival for each time interval. A framework for modelling survival with the application of fuzzy inference system and back-propagation neural network was developed and is described in this thesis. In this framework, a different way of partitioning the input space can be selected to define the membership functions for examples using expert knowledge, equaliser partitioning, fuzzy c-means or subtractive clustering techniques. Further, the rule base can be established by enumerating all possible combinations of membership functions of all inputs. After the initialisation of the fuzzy inference structure, the replication data (until time to event) will be subject to training using the gradient descent and nonnegative least square algorithm to estimate the conditional event probability. This framework is validated over a synthetic dataset and a novel dataset of patients following operative surgery of ovarian cancer. The proposed framework can be applied to estimate the hazard and survival curve between different prognostic factors and survival time with the explanation capabilities.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available