An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data

Lu Cheng; Siddharth Ramchandran; Tommi Vatanen; Niina Lietzén; Riitta Lahesmaa; Aki Vehtari; Harri Lähdesmäki

doi:10.1038/s41467-019-09785-8

An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data

Lu Cheng^*
, Siddharth Ramchandran
, Tommi Vatanen
, Niina Lietzén
, Riitta Lahesmaa
, Aki Vehtari
, Harri Lähdesmäki

^*Corresponding author for this work

Cardiff University
Åbo Akademi University
University of Auckland
Broad Institute

Research output: Contribution to journal › Article › Scientific › peer-review

100 Citations (Scopus)

263 Downloads (Pure)

Abstract

Biomedical research typically involves longitudinal study designs where samples from individuals are measured repeatedly over time and the goal is to identify risk factors (covariates) that are associated with an outcome value. General linear mixed effect models are the standard workhorse for statistical analysis of longitudinal data. However, analysis of longitudinal data can be complicated for reasons such as difficulties in modelling correlated outcome values, functional (time-varying) covariates, nonlinear and non-stationary effects, and model inference. We present LonGP, an additive Gaussian process regression model that is specifically designed for statistical analysis of longitudinal data, which solves these commonly faced challenges. LonGP can model time-varying random effects and non-stationary signals, incorporate multiple kernel learning, and provide interpretable results for the effects of individual covariates and their interactions. We demonstrate LonGP’s performance and accuracy by analysing various simulated and real longitudinal -omics datasets.

Original language	English
Article number	1798
Pages (from-to)	1-11
Number of pages	11
Journal	Nature Communications
Volume	10
Issue number	1
DOIs	https://doi.org/10.1038/s41467-019-09785-8
Publication status	Published - 17 Apr 2019
MoE publication type	A1 Journal article-refereed

Funding

We would like to acknowledge the computational resources provided by the Aalto Science-IT and CSC-IT Center for Science, Finland. This work has been supported by the Academy of Finland Centre of Excellence in Molecular Systems Immunology and Physiology Research 2012-2017 grant 250114; the Academy of Finland grants no. 292660, 292335, 294337, 292482, 287423; JDRF grant no. 17-2013-533, 1-SRA-2017-357-Q-R; the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 663830; and the Business Finland.

Keywords

OUT CROSS-VALIDATION
INFERENCE

Access to Document

10.1038/s41467-019-09785-8Licence: CC BY

s41467-019-09785-8Final published version, 777 KBLicence: CC BY

Cite this

@article{99995bce50034e8a85f2f769a04e0161,

title = "An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data",

abstract = "Biomedical research typically involves longitudinal study designs where samples from individuals are measured repeatedly over time and the goal is to identify risk factors (covariates) that are associated with an outcome value. General linear mixed effect models are the standard workhorse for statistical analysis of longitudinal data. However, analysis of longitudinal data can be complicated for reasons such as difficulties in modelling correlated outcome values, functional (time-varying) covariates, nonlinear and non-stationary effects, and model inference. We present LonGP, an additive Gaussian process regression model that is specifically designed for statistical analysis of longitudinal data, which solves these commonly faced challenges. LonGP can model time-varying random effects and non-stationary signals, incorporate multiple kernel learning, and provide interpretable results for the effects of individual covariates and their interactions. We demonstrate LonGP{\textquoteright}s performance and accuracy by analysing various simulated and real longitudinal -omics datasets.",

keywords = "OUT CROSS-VALIDATION, INFERENCE",

author = "Lu Cheng and Siddharth Ramchandran and Tommi Vatanen and Niina Lietz{\'e}n and Riitta Lahesmaa and Aki Vehtari and Harri L{\"a}hdesm{\"a}ki",

note = "| openaire: EC/H2020/663830/EU//SIRCIW ",

year = "2019",

month = apr,

day = "17",

doi = "10.1038/s41467-019-09785-8",

language = "English",

volume = "10",

pages = "1--11",

journal = "Nature Communications",

issn = "2041-1723",

publisher = "Nature Publishing Group",

number = "1",

}

TY - JOUR

T1 - An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data

AU - Cheng, Lu

AU - Ramchandran, Siddharth

AU - Vatanen, Tommi

AU - Lietzén, Niina

AU - Lahesmaa, Riitta

AU - Vehtari, Aki

AU - Lähdesmäki, Harri

N1 - | openaire: EC/H2020/663830/EU//SIRCIW

PY - 2019/4/17

Y1 - 2019/4/17

N2 - Biomedical research typically involves longitudinal study designs where samples from individuals are measured repeatedly over time and the goal is to identify risk factors (covariates) that are associated with an outcome value. General linear mixed effect models are the standard workhorse for statistical analysis of longitudinal data. However, analysis of longitudinal data can be complicated for reasons such as difficulties in modelling correlated outcome values, functional (time-varying) covariates, nonlinear and non-stationary effects, and model inference. We present LonGP, an additive Gaussian process regression model that is specifically designed for statistical analysis of longitudinal data, which solves these commonly faced challenges. LonGP can model time-varying random effects and non-stationary signals, incorporate multiple kernel learning, and provide interpretable results for the effects of individual covariates and their interactions. We demonstrate LonGP’s performance and accuracy by analysing various simulated and real longitudinal -omics datasets.

AB - Biomedical research typically involves longitudinal study designs where samples from individuals are measured repeatedly over time and the goal is to identify risk factors (covariates) that are associated with an outcome value. General linear mixed effect models are the standard workhorse for statistical analysis of longitudinal data. However, analysis of longitudinal data can be complicated for reasons such as difficulties in modelling correlated outcome values, functional (time-varying) covariates, nonlinear and non-stationary effects, and model inference. We present LonGP, an additive Gaussian process regression model that is specifically designed for statistical analysis of longitudinal data, which solves these commonly faced challenges. LonGP can model time-varying random effects and non-stationary signals, incorporate multiple kernel learning, and provide interpretable results for the effects of individual covariates and their interactions. We demonstrate LonGP’s performance and accuracy by analysing various simulated and real longitudinal -omics datasets.

KW - OUT CROSS-VALIDATION

KW - INFERENCE

UR - https://www.scopus.com/pages/publications/85064561361

U2 - 10.1038/s41467-019-09785-8

DO - 10.1038/s41467-019-09785-8

M3 - Article

AN - SCOPUS:85064561361

SN - 2041-1723

VL - 10

SP - 1

EP - 11

JO - Nature Communications

JF - Nature Communications

IS - 1

M1 - 1798

ER -

An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Science-IT

Cite this

An additive Gaussian process regression model for interpretable non-parametric analysis of longitudinal data

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Equipment

Science-IT

Cite this