Please use this identifier to cite or link to this item:
Title: From automated to data-driven large-scale dietary assessment
Authors: Osadchiy, Timur
Issue Date: 2020
Publisher: Newcastle University
Abstract: Dietary assessment surveys are an important tool for measuring and/or monitoring the nutritional profile of a population. The analysis of data that is collected in these surveys helps to develop health care guidelines and policies that minimise the risk of diet related diseases on a national scale. For years these surveys had to be conducted in a form of an interview by trained researchers with a nutritional background. The emergence of systems that automate interviewer-led protocols and transform these interviews into online surveys has addressed financial limitations and brought scalability into dietary assessment studies. In the meantime, online dietary assessment surveys mostly copy the interviewer-led procedures and inherit some of their methodological issues that lead to misreporting of dietary intake and lower the accuracy of assessment. This thesis primarily focuses on the issues related to human-memory, motivation of respondents to take part in dietary assessment studies, and the usability of survey interfaces. This work pinpoints the elements of automated dietary assessment systems, where these issues affect the accuracy of results. This analysis is then translated into three research questions of this thesis. Challenges related to human-memory are then addressed by developing and evaluating a recommender system for prompting omitted foods in online dietary assessment surveys. This work also explores short retention intervals (i.e. time between an intake and recall) as another method for recall assistance. As a way to motivate respondents to take part in dietary assessment surveys this thesis explores tailored dietary feedback provided to respondents at the end of a survey. Usability and performance of new methods are analysed in real-life dietary assessment surveys using a usability framework developed for this research. Acceptance of the methods is analysed using thematic analysis of transcribed interviews with respondents. Research activities conducted during this work provide some support to hypotheses defined in the research questions.
Description: Ph. D. Thesis.
Appears in Collections:School of Computing Science

Files in This Item:
File Description SizeFormat 
Osadchiy T 2020.pdfThesis12.56 MBAdobe PDFView/Open
dspacelicence.pdfLicence43.82 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.