Use this URL to cite or link to this record in EThOS:
Title: An investigation into neuropsychological profiles in anorexia nervosa and associated clinical and demographic variables
Author: Drake, A. C. L.
ISNI:       0000 0004 7969 5676
Awarding Body: University of Exeter
Current Institution: University of Exeter
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Objective: Treatment outcomes for anorexia nervosa (AN) remain unsatisfactory. Substantial research has investigated the neuropsychological effects of AN, often with mixed results. One explanation for the inconsistencies is that there exist several distinct neuropsychological profiles within AN. Profiles have been reported, though not associated with clinical or demographic variables, limiting their utility. Suboptimal statistical techniques may undermine these findings. Method: An existing dataset of healthy controls (HCs) and AN patients (n = 423) was subjected to secondary analysis using latent profile analysis and a neural network to investigate latent profiles and the existence of non-linear neuropsychological structure. Profiles were compared with respect to demographic and clinical variables. Results: The latent profile analysis revealed five AN neuropsychological profiles. Patients in a globally neuropsychologically impaired profile were older than those in a high-average with high verbal profile and weighed less than those in an average performance profile. A non-linear neural network failed to outperform a linear neural network on a diagnosis classification task. Discussion: The five-profile solution extended the neuropsychological groups previously found in the literature. This study is the first to successfully associate latent neuropsychological profile to clinically meaningful variables, though the profile in which differences were observed was tiny (7% of patients). None of the discovered profiles differed in terms of anxiety, undermining support for the noradrenergic hypothesis of AN. The failure of the non-linear neural network to outperform the linear network indicates that AN neuropsychological ability does not contain significant non-linearity, indicating that conventional statistical techniques can model them.
Supervisor: Frampton, I. ; Karl, A. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: anorexia nervosa ; neuropsychology ; systematic review ; latent profile analysis ; machine learning