Sabate Barbera, David;
(2024)
Design and Implementation of a Novel Spoken Utterance Verification System for Anomia Treatment.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
Sabate Barbera_10202755_thesis.pdf Download (8MB) | Preview |
Abstract
This thesis explores the design, implementation and evaluation of two automatic speech recognition (ASR) systems: 1. a naming utterance verifier for aphasia (NUVA) for deployment into the app iTALKbetter to treat anomia in chronic post-stroke aphasia; 2. a proper naming utterance verifier for dementia (PNUVD) for deployment into the app Gotcha! to treat proper name anomia in early-stage dementia. In the methods section, I present a new framework, 'Neural Inference’, required for deploying deep learning models on mobile devices. I also propose a novel probabilistic metric based on interrater reliability theory to explore evaluation for clinical applications. In Chapter 3, NUVA’s performance is evaluated on curated data collected from a pre-therapy block of the clinical trial for iTALKbetter (n=8). Results show that NUVA naming assessment accuracy is equal or superior to a commercial baseline and at par with speech and language therapists (SLTs) in two of the participants, ranging between Moderate and Strong agreement with SLTs. In Chapter 4, NUVA’s performance is evaluated on realistic therapeutic conditions using participants’ data who completed the therapy block (n=8) of the iTALKbetter clinical trial. Results show that NUVA naming assessment reliability is consistently superior or equal to the commercial baseline, except for one outlier, ranging between Weak and Strong with the rater. In Chapter 5, PNUVD’s performance is evaluated on realistic therapeutic conditions using participants’ data who completed the therapy block (n=10) of the Gotcha! clinical trial. Results show that both PNUVD and the commercial baseline struggle with proper naming assessment in this intervention. However, only PNUVD reaches Moderate agreement with the rater in 4 participants. In the general discussion chapter, I discuss the advantages and limitations of NUVA and PNUVD and suggest possible improvements and future avenues of research.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Design and Implementation of a Novel Spoken Utterance Verification System for Anomia Treatment |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2025. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Brain Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Brain Sciences > Div of Psychology and Lang Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Brain Sciences > Div of Psychology and Lang Sciences > Institute of Cognitive Neuroscience UCL |
| URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10202755 |
Archive Staff Only
 |
View Item |
