Profiling Movement Behaviours in Pre-School Children: A Self-Organized Map Approach

Cain Clark, Michael Duncan, Emma Eyre, Gareth Stratton, Xavier García-Massó, Isaac Estevan

Research output: Contribution to journalArticle

Abstract

Application of machine learning techniques has the potential to yield unseen insights into movement and permits visualisation of complex behaviours and tangible profiles. The aim of this study was to identify profiles of relative motor competence (MC) and movement behaviours in pre-school children using novel analytics. One-hundred and twenty-five children (4.3±0.5y, 1.04±0.05m, 17.8±3.2kg, BMI: 16.2±1.9 kg.m2) took part in this study. Measures included accelerometer-derived 24-h activity, MC (Movement Assessment Battery for Children 2nd edition), height, weight and waist circumference, from which zBMI was derived. Self-Organized Map (SOM) analysis was used to classify participants’ profiles and a k-means cluster analysis was used to classify the neurons into larger groups according to the input variables. These clusters were used to describe the individuals’ characteristics according to their MC and PA compositions. The SOM analysis indicated five profiles according to MC and PA. One cluster was identified as having both the lowest MC and MVPA (profile 2), whilst profiles 4 and 5 show moderate-high values of PA and MC. We present a novel pathway to profiling complex tenets of human movement and behaviour, which has never previously been implemented in pre-school children, highlighting that the focus should change from obesity monitoring, to ‘moving well’.
Original languageEnglish
Pages (from-to)(In-press)
JournalJournal of Sports Sciences
Volume(In-press)
Early online date7 Nov 2019
Publication statusE-pub ahead of print - 7 Nov 2019

Fingerprint

Mental Competency
Waist Circumference
Cluster Analysis
Motor Activity
Obesity
Neurons
Weights and Measures

Keywords

  • Motor Competence
  • Machine Learning
  • Unsupervised
  • Cluster Analysis
  • Physical Activity

Cite this

Profiling Movement Behaviours in Pre-School Children: A Self-Organized Map Approach. / Clark, Cain; Duncan, Michael; Eyre, Emma; Stratton, Gareth; García-Massó, Xavier ; Estevan, Isaac.

In: Journal of Sports Sciences, Vol. (In-press), 07.11.2019, p. (In-press).

Research output: Contribution to journalArticle

Clark, Cain ; Duncan, Michael ; Eyre, Emma ; Stratton, Gareth ; García-Massó, Xavier ; Estevan, Isaac. / Profiling Movement Behaviours in Pre-School Children: A Self-Organized Map Approach. In: Journal of Sports Sciences. 2019 ; Vol. (In-press). pp. (In-press).
@article{013ac33df8154107a27e7dd3bd249a44,
title = "Profiling Movement Behaviours in Pre-School Children: A Self-Organized Map Approach",
abstract = "Application of machine learning techniques has the potential to yield unseen insights into movement and permits visualisation of complex behaviours and tangible profiles. The aim of this study was to identify profiles of relative motor competence (MC) and movement behaviours in pre-school children using novel analytics. One-hundred and twenty-five children (4.3±0.5y, 1.04±0.05m, 17.8±3.2kg, BMI: 16.2±1.9 kg.m2) took part in this study. Measures included accelerometer-derived 24-h activity, MC (Movement Assessment Battery for Children 2nd edition), height, weight and waist circumference, from which zBMI was derived. Self-Organized Map (SOM) analysis was used to classify participants’ profiles and a k-means cluster analysis was used to classify the neurons into larger groups according to the input variables. These clusters were used to describe the individuals’ characteristics according to their MC and PA compositions. The SOM analysis indicated five profiles according to MC and PA. One cluster was identified as having both the lowest MC and MVPA (profile 2), whilst profiles 4 and 5 show moderate-high values of PA and MC. We present a novel pathway to profiling complex tenets of human movement and behaviour, which has never previously been implemented in pre-school children, highlighting that the focus should change from obesity monitoring, to ‘moving well’.",
keywords = "Motor Competence, Machine Learning, Unsupervised, Cluster Analysis, Physical Activity",
author = "Cain Clark and Michael Duncan and Emma Eyre and Gareth Stratton and Xavier Garc{\'i}a-Mass{\'o} and Isaac Estevan",
year = "2019",
month = "11",
day = "7",
language = "English",
volume = "(In-press)",
pages = "(In--press)",
journal = "Journal of Sports Sciences",
issn = "0264-0414",
publisher = "Taylor & Francis",

}

TY - JOUR

T1 - Profiling Movement Behaviours in Pre-School Children: A Self-Organized Map Approach

AU - Clark, Cain

AU - Duncan, Michael

AU - Eyre, Emma

AU - Stratton, Gareth

AU - García-Massó, Xavier

AU - Estevan, Isaac

PY - 2019/11/7

Y1 - 2019/11/7

N2 - Application of machine learning techniques has the potential to yield unseen insights into movement and permits visualisation of complex behaviours and tangible profiles. The aim of this study was to identify profiles of relative motor competence (MC) and movement behaviours in pre-school children using novel analytics. One-hundred and twenty-five children (4.3±0.5y, 1.04±0.05m, 17.8±3.2kg, BMI: 16.2±1.9 kg.m2) took part in this study. Measures included accelerometer-derived 24-h activity, MC (Movement Assessment Battery for Children 2nd edition), height, weight and waist circumference, from which zBMI was derived. Self-Organized Map (SOM) analysis was used to classify participants’ profiles and a k-means cluster analysis was used to classify the neurons into larger groups according to the input variables. These clusters were used to describe the individuals’ characteristics according to their MC and PA compositions. The SOM analysis indicated five profiles according to MC and PA. One cluster was identified as having both the lowest MC and MVPA (profile 2), whilst profiles 4 and 5 show moderate-high values of PA and MC. We present a novel pathway to profiling complex tenets of human movement and behaviour, which has never previously been implemented in pre-school children, highlighting that the focus should change from obesity monitoring, to ‘moving well’.

AB - Application of machine learning techniques has the potential to yield unseen insights into movement and permits visualisation of complex behaviours and tangible profiles. The aim of this study was to identify profiles of relative motor competence (MC) and movement behaviours in pre-school children using novel analytics. One-hundred and twenty-five children (4.3±0.5y, 1.04±0.05m, 17.8±3.2kg, BMI: 16.2±1.9 kg.m2) took part in this study. Measures included accelerometer-derived 24-h activity, MC (Movement Assessment Battery for Children 2nd edition), height, weight and waist circumference, from which zBMI was derived. Self-Organized Map (SOM) analysis was used to classify participants’ profiles and a k-means cluster analysis was used to classify the neurons into larger groups according to the input variables. These clusters were used to describe the individuals’ characteristics according to their MC and PA compositions. The SOM analysis indicated five profiles according to MC and PA. One cluster was identified as having both the lowest MC and MVPA (profile 2), whilst profiles 4 and 5 show moderate-high values of PA and MC. We present a novel pathway to profiling complex tenets of human movement and behaviour, which has never previously been implemented in pre-school children, highlighting that the focus should change from obesity monitoring, to ‘moving well’.

KW - Motor Competence

KW - Machine Learning

KW - Unsupervised

KW - Cluster Analysis

KW - Physical Activity

M3 - Article

VL - (In-press)

SP - (In-press)

JO - Journal of Sports Sciences

JF - Journal of Sports Sciences

SN - 0264-0414

ER -