A digital technology framework to optimise the self-management of obesity

Patrick McAllister, Huiru Zheng, Raymond Bond, Anne Moorhead

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Obesity is increasing globally and can cause major chronic conditions. Much research has been completed in utilising digital technologies to optimise the self-management of obesity. This research proposes an obesity management framework which highlights digital technologies to promote self-management of obesity. This work discusses preliminary research using image classification to promote food logging and crowdsourcing to determine calorie content of food images through aggregating the predictions of experts and non-experts. Preliminary results from image classification show SMO classifier achieved 73.87% accuracy in classifying 15 food items, which is promising as computer vision methods could be incorporated into food logging methods. Crowdsourcing results show that aggregated expert group mode percentage error was +2.60% (SD 3.87) in predicting calories in meals and non-expert group mode percentage error was +29.07% (SD 20.48). Further analysis on the crowdsourcing dataset will be completed to ascertain how many experts or nonexperts is needed to get the most accurate calorie prediction.
LanguageEnglish
Title of host publicationUnknown Host Publication
Pages1126-1131
Number of pages5
DOIs
Publication statusAccepted/In press - 27 Jul 2016
EventACM International Joint Conference on Pervasive and Ubiquitous Computing - Germany
Duration: 27 Jul 2016 → …

Conference

ConferenceACM International Joint Conference on Pervasive and Ubiquitous Computing
Period27/07/16 → …

Fingerprint

Image classification
Computer vision
Classifiers
Crowdsourcing

Keywords

  • Obesity
  • Self-Management
  • Connected Health
  • Mobile Computing
  • Machine Learning
  • Machine Vision
  • Crowdsourcing

Cite this

@inproceedings{45af919d8075469aba16c4b22d40b74b,
title = "A digital technology framework to optimise the self-management of obesity",
abstract = "Obesity is increasing globally and can cause major chronic conditions. Much research has been completed in utilising digital technologies to optimise the self-management of obesity. This research proposes an obesity management framework which highlights digital technologies to promote self-management of obesity. This work discusses preliminary research using image classification to promote food logging and crowdsourcing to determine calorie content of food images through aggregating the predictions of experts and non-experts. Preliminary results from image classification show SMO classifier achieved 73.87{\%} accuracy in classifying 15 food items, which is promising as computer vision methods could be incorporated into food logging methods. Crowdsourcing results show that aggregated expert group mode percentage error was +2.60{\%} (SD 3.87) in predicting calories in meals and non-expert group mode percentage error was +29.07{\%} (SD 20.48). Further analysis on the crowdsourcing dataset will be completed to ascertain how many experts or nonexperts is needed to get the most accurate calorie prediction.",
keywords = "Obesity, Self-Management, Connected Health, Mobile Computing, Machine Learning, Machine Vision, Crowdsourcing",
author = "Patrick McAllister and Huiru Zheng and Raymond Bond and Anne Moorhead",
year = "2016",
month = "7",
day = "27",
doi = "10.1145/2968219.2978096",
language = "English",
isbn = "978-1-4503-4462-3",
pages = "1126--1131",
booktitle = "Unknown Host Publication",

}

McAllister, P, Zheng, H, Bond, R & Moorhead, A 2016, A digital technology framework to optimise the self-management of obesity. in Unknown Host Publication. pp. 1126-1131, ACM International Joint Conference on Pervasive and Ubiquitous Computing, 27/07/16. https://doi.org/10.1145/2968219.2978096

A digital technology framework to optimise the self-management of obesity. / McAllister, Patrick; Zheng, Huiru; Bond, Raymond; Moorhead, Anne.

Unknown Host Publication. 2016. p. 1126-1131.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A digital technology framework to optimise the self-management of obesity

AU - McAllister, Patrick

AU - Zheng, Huiru

AU - Bond, Raymond

AU - Moorhead, Anne

PY - 2016/7/27

Y1 - 2016/7/27

N2 - Obesity is increasing globally and can cause major chronic conditions. Much research has been completed in utilising digital technologies to optimise the self-management of obesity. This research proposes an obesity management framework which highlights digital technologies to promote self-management of obesity. This work discusses preliminary research using image classification to promote food logging and crowdsourcing to determine calorie content of food images through aggregating the predictions of experts and non-experts. Preliminary results from image classification show SMO classifier achieved 73.87% accuracy in classifying 15 food items, which is promising as computer vision methods could be incorporated into food logging methods. Crowdsourcing results show that aggregated expert group mode percentage error was +2.60% (SD 3.87) in predicting calories in meals and non-expert group mode percentage error was +29.07% (SD 20.48). Further analysis on the crowdsourcing dataset will be completed to ascertain how many experts or nonexperts is needed to get the most accurate calorie prediction.

AB - Obesity is increasing globally and can cause major chronic conditions. Much research has been completed in utilising digital technologies to optimise the self-management of obesity. This research proposes an obesity management framework which highlights digital technologies to promote self-management of obesity. This work discusses preliminary research using image classification to promote food logging and crowdsourcing to determine calorie content of food images through aggregating the predictions of experts and non-experts. Preliminary results from image classification show SMO classifier achieved 73.87% accuracy in classifying 15 food items, which is promising as computer vision methods could be incorporated into food logging methods. Crowdsourcing results show that aggregated expert group mode percentage error was +2.60% (SD 3.87) in predicting calories in meals and non-expert group mode percentage error was +29.07% (SD 20.48). Further analysis on the crowdsourcing dataset will be completed to ascertain how many experts or nonexperts is needed to get the most accurate calorie prediction.

KW - Obesity

KW - Self-Management

KW - Connected Health

KW - Mobile Computing

KW - Machine Learning

KW - Machine Vision

KW - Crowdsourcing

U2 - 10.1145/2968219.2978096

DO - 10.1145/2968219.2978096

M3 - Conference contribution

SN - 978-1-4503-4462-3

SP - 1126

EP - 1131

BT - Unknown Host Publication

ER -