Dynamic child growth prediction

A comparative methods approach

Andrada Ivanescu, Ciprian M. Crainiceanu, William Checkley

Research output: Contribution to journalArticleResearchpeer-review

Abstract

We introduce a class of dynamic regression models designed to predict the future of growth curves based on their historical dynamics. This class of models incorporates both baseline and time-dependent covariates, start with simple regression models and build up to dynamic function-on-function regressions. We compare the performance of the dynamic prediction models in a variety of signal-to-noise scenarios and provide practical solutions for model selection. We conclude that (a) prediction performance increases substantially when using the entire growth history relative to using only the last and first observation; (b) smoothing incorporated using functional regression approaches increases prediction performance; and (c) the interpretation of model parameters is substantially improved using functional regression approaches. Because many growth curve datasets exhibit missing and noisy data, we propose a bootstrap of subjects approach to account for the variability associated with the missing data imputation and smoothing. Methods are motivated by and applied to the CONTENT dataset, a study that collected monthly child growth data on 197 children from birth until month 15. R code describing the fitting approaches is provided in a supplementary file.

Original languageEnglish
Pages (from-to)468-493
Number of pages26
JournalStatistical Modelling
Volume17
Issue number6
DOIs
StatePublished - 1 Dec 2017

Fingerprint

Growth Curve
Performance Prediction
Missing Data
Smoothing
Regression Model
Dynamic Model
Regression
Time-dependent Covariates
Prediction
Imputation
Noisy Data
Regression Function
Model Selection
Prediction Model
Bootstrap
Baseline
Entire
Predict
Scenarios
Model

Keywords

  • Function-on-function regression
  • functional data
  • functional regression
  • height
  • longitudinal data
  • weight

Cite this

Ivanescu, Andrada ; Crainiceanu, Ciprian M. ; Checkley, William. / Dynamic child growth prediction : A comparative methods approach. In: Statistical Modelling. 2017 ; Vol. 17, No. 6. pp. 468-493.
@article{33eea27153d341cf900bb5dcdf7c677c,
title = "Dynamic child growth prediction: A comparative methods approach",
abstract = "We introduce a class of dynamic regression models designed to predict the future of growth curves based on their historical dynamics. This class of models incorporates both baseline and time-dependent covariates, start with simple regression models and build up to dynamic function-on-function regressions. We compare the performance of the dynamic prediction models in a variety of signal-to-noise scenarios and provide practical solutions for model selection. We conclude that (a) prediction performance increases substantially when using the entire growth history relative to using only the last and first observation; (b) smoothing incorporated using functional regression approaches increases prediction performance; and (c) the interpretation of model parameters is substantially improved using functional regression approaches. Because many growth curve datasets exhibit missing and noisy data, we propose a bootstrap of subjects approach to account for the variability associated with the missing data imputation and smoothing. Methods are motivated by and applied to the CONTENT dataset, a study that collected monthly child growth data on 197 children from birth until month 15. R code describing the fitting approaches is provided in a supplementary file.",
keywords = "Function-on-function regression, functional data, functional regression, height, longitudinal data, weight",
author = "Andrada Ivanescu and Crainiceanu, {Ciprian M.} and William Checkley",
year = "2017",
month = "12",
day = "1",
doi = "10.1177/1471082X17707619",
language = "English",
volume = "17",
pages = "468--493",
journal = "Statistical Modelling",
issn = "1471-082X",
publisher = "SAGE Publications Ltd",
number = "6",

}

Dynamic child growth prediction : A comparative methods approach. / Ivanescu, Andrada; Crainiceanu, Ciprian M.; Checkley, William.

In: Statistical Modelling, Vol. 17, No. 6, 01.12.2017, p. 468-493.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Dynamic child growth prediction

T2 - A comparative methods approach

AU - Ivanescu, Andrada

AU - Crainiceanu, Ciprian M.

AU - Checkley, William

PY - 2017/12/1

Y1 - 2017/12/1

N2 - We introduce a class of dynamic regression models designed to predict the future of growth curves based on their historical dynamics. This class of models incorporates both baseline and time-dependent covariates, start with simple regression models and build up to dynamic function-on-function regressions. We compare the performance of the dynamic prediction models in a variety of signal-to-noise scenarios and provide practical solutions for model selection. We conclude that (a) prediction performance increases substantially when using the entire growth history relative to using only the last and first observation; (b) smoothing incorporated using functional regression approaches increases prediction performance; and (c) the interpretation of model parameters is substantially improved using functional regression approaches. Because many growth curve datasets exhibit missing and noisy data, we propose a bootstrap of subjects approach to account for the variability associated with the missing data imputation and smoothing. Methods are motivated by and applied to the CONTENT dataset, a study that collected monthly child growth data on 197 children from birth until month 15. R code describing the fitting approaches is provided in a supplementary file.

AB - We introduce a class of dynamic regression models designed to predict the future of growth curves based on their historical dynamics. This class of models incorporates both baseline and time-dependent covariates, start with simple regression models and build up to dynamic function-on-function regressions. We compare the performance of the dynamic prediction models in a variety of signal-to-noise scenarios and provide practical solutions for model selection. We conclude that (a) prediction performance increases substantially when using the entire growth history relative to using only the last and first observation; (b) smoothing incorporated using functional regression approaches increases prediction performance; and (c) the interpretation of model parameters is substantially improved using functional regression approaches. Because many growth curve datasets exhibit missing and noisy data, we propose a bootstrap of subjects approach to account for the variability associated with the missing data imputation and smoothing. Methods are motivated by and applied to the CONTENT dataset, a study that collected monthly child growth data on 197 children from birth until month 15. R code describing the fitting approaches is provided in a supplementary file.

KW - Function-on-function regression

KW - functional data

KW - functional regression

KW - height

KW - longitudinal data

KW - weight

UR - http://www.scopus.com/inward/record.url?scp=85033486591&partnerID=8YFLogxK

U2 - 10.1177/1471082X17707619

DO - 10.1177/1471082X17707619

M3 - Article

VL - 17

SP - 468

EP - 493

JO - Statistical Modelling

JF - Statistical Modelling

SN - 1471-082X

IS - 6

ER -