Structural differences in adolescent brainscan predict alcohol misuse

IMAGEN Consortium

doi:10.7554/eLife.77545

Structural differences in adolescent brainscan predict alcohol misuse

IMAGEN Consortium

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

11 Citations (Scopus)

Abstract

Alcohol misuse during adolescence (AAM) has been associated with disruptive development of adolescent brains. In this longitudinal machine learning (ML) study, we could predict AAM significantly from brain structure (T1-weighted imaging and DTI) with accuracies of 73 − 78% in the IMAGEN dataset (n ∼ 1182). Our results not only show that structural differences in brain can predict AAM, but also suggests that such differences might precede AAM behavior in the data. We predicted ten phenotypes of AAM at age 22 using brain MRI features at ages 14, 19, and 22. Binge drinking was found to be the most predictable phenotype. The most informative brain features were located in the ventricular CSF, and in white matter tracts of the corpus callosum, internal capsule, and brain stem. In the cortex, they were spread across the occipital, frontal, and temporal lobes and in the cingulate cortex. We also experimented with four different ML models and several confound control techniques. Support Vector Machine (SVM) with rbf kernel and Gradient Boosting consistently performed better than the linear models, linear SVM and Logistic Regression. Our study also demonstrates how the choice of the predicted phenotype, ML model, and confound correction technique are all crucial decisions in an explorative ML study analyzing psychiatric disorders with small effect sizes such as AAM.

Original language	English
Article number	e77545
Journal	eLife
Volume	11
DOIs	https://doi.org/10.7554/eLife.77545
Publication status	Published - 26 May 2022

Keywords

adolescence alcohol misuse
alcohol use disorder
computational biology
confound control
data science for psychiatry
machine learning
magnetic resonance imaging
multivariate analysis
neuroscience
psychiatric research
systems biology

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.7554/eLife.77545

Cite this

@article{5633e75d262e485c94f207bd02689e06,

title = "Structural differences in adolescent brainscan predict alcohol misuse",

abstract = "Alcohol misuse during adolescence (AAM) has been associated with disruptive development of adolescent brains. In this longitudinal machine learning (ML) study, we could predict AAM significantly from brain structure (T1-weighted imaging and DTI) with accuracies of 73 − 78% in the IMAGEN dataset (n ∼ 1182). Our results not only show that structural differences in brain can predict AAM, but also suggests that such differences might precede AAM behavior in the data. We predicted ten phenotypes of AAM at age 22 using brain MRI features at ages 14, 19, and 22. Binge drinking was found to be the most predictable phenotype. The most informative brain features were located in the ventricular CSF, and in white matter tracts of the corpus callosum, internal capsule, and brain stem. In the cortex, they were spread across the occipital, frontal, and temporal lobes and in the cingulate cortex. We also experimented with four different ML models and several confound control techniques. Support Vector Machine (SVM) with rbf kernel and Gradient Boosting consistently performed better than the linear models, linear SVM and Logistic Regression. Our study also demonstrates how the choice of the predicted phenotype, ML model, and confound correction technique are all crucial decisions in an explorative ML study analyzing psychiatric disorders with small effect sizes such as AAM.",

keywords = "adolescence alcohol misuse, alcohol use disorder, computational biology, confound control, data science for psychiatry, machine learning, magnetic resonance imaging, multivariate analysis, neuroscience, psychiatric research, systems biology",

author = "{IMAGEN Consortium} and Rane, {Roshan Prakash} and {de Man}, {Evert Ferdinand} and Kim, {Ji Hoon} and Kai G{\"o}rgen and Mira Tschorn and Rapp, {Michael A.} and Tobias Banaschewski and Bokde, {Arun L.W.} and Sylvane Desrivieres and Herta Flor and Antoine Grigis and Hugh Garavan and Gowland, {Penny A.} and R{\"u}diger Br{\"u}hl and Martinot, {Jean Luc} and Martinot, {Marie Laure Paillere} and Eric Artiges and Frauke Nees and {Papadopoulos Orfanos}, Dimitri and Herve Lemaitre and Tomas Paus and Luise Poustka and Juliane Fr{\"o}hner and Lauren Robinson and Smolka, {Michael N.} and Jeanne Winterer and Robert Whelan and Gunter Schumann and Henrik Walter and Andreas Heinz and Kerstin Ritter",

note = "Funding Information: 548 We acknowledge support from the German Research Foundation (DFG, 389563835; 402170461-TRR 265; 549 414984028-CRC 1404; EXC 2002/1 “Science of Intelligence” – project number 390523135), the Brain 550 & Behavior Research Foundation (NARSAD grant) and the Manfred and Ursula-M{\"u}ller Stiftung. Gunter 551 Schumann is a recipient of an Alexander von Humboldt Preis and a National Science Foundation of China 552 (NSFC) Research Fund for International Scientists (Grant No. 82150710554). Publisher Copyright: {\textcopyright} 2022, eLife Sciences Publications Ltd. All rights reserved.",

year = "2022",

month = may,

day = "26",

doi = "10.7554/eLife.77545",

language = "English",

volume = "11",

journal = "eLife",

issn = "2050-084X",

publisher = "eLife Sciences Publications, Ltd",

}

TY - JOUR

T1 - Structural differences in adolescent brainscan predict alcohol misuse

AU - IMAGEN Consortium

AU - Rane, Roshan Prakash

AU - de Man, Evert Ferdinand

AU - Kim, Ji Hoon

AU - Görgen, Kai

AU - Tschorn, Mira

AU - Rapp, Michael A.

AU - Banaschewski, Tobias

AU - Bokde, Arun L.W.

AU - Desrivieres, Sylvane

AU - Flor, Herta

AU - Grigis, Antoine

AU - Garavan, Hugh

AU - Gowland, Penny A.

AU - Brühl, Rüdiger

AU - Martinot, Jean Luc

AU - Martinot, Marie Laure Paillere

AU - Artiges, Eric

AU - Nees, Frauke

AU - Papadopoulos Orfanos, Dimitri

AU - Lemaitre, Herve

AU - Paus, Tomas

AU - Poustka, Luise

AU - Fröhner, Juliane

AU - Robinson, Lauren

AU - Smolka, Michael N.

AU - Winterer, Jeanne

AU - Whelan, Robert

AU - Schumann, Gunter

AU - Walter, Henrik

AU - Heinz, Andreas

AU - Ritter, Kerstin

N1 - Funding Information: 548 We acknowledge support from the German Research Foundation (DFG, 389563835; 402170461-TRR 265; 549 414984028-CRC 1404; EXC 2002/1 “Science of Intelligence” – project number 390523135), the Brain 550 & Behavior Research Foundation (NARSAD grant) and the Manfred and Ursula-Müller Stiftung. Gunter 551 Schumann is a recipient of an Alexander von Humboldt Preis and a National Science Foundation of China 552 (NSFC) Research Fund for International Scientists (Grant No. 82150710554). Publisher Copyright: © 2022, eLife Sciences Publications Ltd. All rights reserved.

PY - 2022/5/26

Y1 - 2022/5/26

N2 - Alcohol misuse during adolescence (AAM) has been associated with disruptive development of adolescent brains. In this longitudinal machine learning (ML) study, we could predict AAM significantly from brain structure (T1-weighted imaging and DTI) with accuracies of 73 − 78% in the IMAGEN dataset (n ∼ 1182). Our results not only show that structural differences in brain can predict AAM, but also suggests that such differences might precede AAM behavior in the data. We predicted ten phenotypes of AAM at age 22 using brain MRI features at ages 14, 19, and 22. Binge drinking was found to be the most predictable phenotype. The most informative brain features were located in the ventricular CSF, and in white matter tracts of the corpus callosum, internal capsule, and brain stem. In the cortex, they were spread across the occipital, frontal, and temporal lobes and in the cingulate cortex. We also experimented with four different ML models and several confound control techniques. Support Vector Machine (SVM) with rbf kernel and Gradient Boosting consistently performed better than the linear models, linear SVM and Logistic Regression. Our study also demonstrates how the choice of the predicted phenotype, ML model, and confound correction technique are all crucial decisions in an explorative ML study analyzing psychiatric disorders with small effect sizes such as AAM.

AB - Alcohol misuse during adolescence (AAM) has been associated with disruptive development of adolescent brains. In this longitudinal machine learning (ML) study, we could predict AAM significantly from brain structure (T1-weighted imaging and DTI) with accuracies of 73 − 78% in the IMAGEN dataset (n ∼ 1182). Our results not only show that structural differences in brain can predict AAM, but also suggests that such differences might precede AAM behavior in the data. We predicted ten phenotypes of AAM at age 22 using brain MRI features at ages 14, 19, and 22. Binge drinking was found to be the most predictable phenotype. The most informative brain features were located in the ventricular CSF, and in white matter tracts of the corpus callosum, internal capsule, and brain stem. In the cortex, they were spread across the occipital, frontal, and temporal lobes and in the cingulate cortex. We also experimented with four different ML models and several confound control techniques. Support Vector Machine (SVM) with rbf kernel and Gradient Boosting consistently performed better than the linear models, linear SVM and Logistic Regression. Our study also demonstrates how the choice of the predicted phenotype, ML model, and confound correction technique are all crucial decisions in an explorative ML study analyzing psychiatric disorders with small effect sizes such as AAM.

KW - adolescence alcohol misuse

KW - alcohol use disorder

KW - computational biology

KW - confound control

KW - data science for psychiatry

KW - machine learning

KW - magnetic resonance imaging

KW - multivariate analysis

KW - neuroscience

KW - psychiatric research

KW - systems biology

UR - http://www.scopus.com/inward/record.url?scp=85133646416&partnerID=8YFLogxK

U2 - 10.7554/eLife.77545

DO - 10.7554/eLife.77545

M3 - Article

C2 - 35616520

AN - SCOPUS:85133646416

SN - 2050-084X

VL - 11

JO - eLife

JF - eLife

M1 - e77545

ER -

Structural differences in adolescent brainscan predict alcohol misuse

Abstract

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this