A proof-of-concept study applying machine learning methods to putative risk factors for eating disorders: Results from the multi-centre European project on healthy eating

TY - JOUR

T1 - A proof-of-concept study applying machine learning methods to putative risk factors for eating disorders

T2 - Results from the multi-centre European project on healthy eating

AU - Krug, I.

AU - Linardon, J.

AU - Greenwood, C.

AU - Youssef, G.

AU - Treasure, J.

AU - Fernandez-Aranda, F.

AU - Karwautz, A.

AU - Wagner, G.

AU - Collier, D.

AU - Anderluh, M.

AU - Tchanturia, K.

AU - Ricca, V.

AU - Sorbi, S.

AU - Nacmias, B.

AU - Bellodi, L.

AU - Fuller-Tyszkiewicz, M.

N1 - Funding Information: Financial support was received from the European Union (Framework – V Multicenter Research Grant, QCK1-1999-916). CIBERobn is an initiative of ISCIII. J.L holds a National Health and Medical Research Council Investigator Grant (APP1196948). Publisher Copyright: © The Author(s), 2021. Published by Cambridge University Press.

PY - 2021/11/29

Y1 - 2021/11/29

N2 - Background Despite a wide range of proposed risk factors and theoretical models, prediction of eating disorder (ED) onset remains poor. This study undertook the first comparison of two machine learning (ML) approaches [penalised logistic regression (LASSO), and prediction rule ensembles (PREs)] to conventional logistic regression (LR) models to enhance prediction of ED onset and differential ED diagnoses from a range of putative risk factors. Method Data were part of a European Project and comprised 1402 participants, 642 ED patients [52% with anorexia nervosa (AN) and 40% with bulimia nervosa (BN)] and 760 controls. The Cross-Cultural Risk Factor Questionnaire, which assesses retrospectively a range of sociocultural and psychological ED risk factors occurring before the age of 12 years (46 predictors in total), was used. Results All three statistical approaches had satisfactory model accuracy, with an average area under the curve (AUC) of 86% for predicting ED onset and 70% for predicting AN v. BN. Predictive performance was greatest for the two regression methods (LR and LASSO), although the PRE technique relied on fewer predictors with comparable accuracy. The individual risk factors differed depending on the outcome classification (EDs v. non-EDs and AN v. BN). Conclusions Even though the conventional LR performed comparably to the ML approaches in terms of predictive accuracy, the ML methods produced more parsimonious predictive models. ML approaches offer a viable way to modify screening practices for ED risk that balance accuracy against participant burden.

AB - Background Despite a wide range of proposed risk factors and theoretical models, prediction of eating disorder (ED) onset remains poor. This study undertook the first comparison of two machine learning (ML) approaches [penalised logistic regression (LASSO), and prediction rule ensembles (PREs)] to conventional logistic regression (LR) models to enhance prediction of ED onset and differential ED diagnoses from a range of putative risk factors. Method Data were part of a European Project and comprised 1402 participants, 642 ED patients [52% with anorexia nervosa (AN) and 40% with bulimia nervosa (BN)] and 760 controls. The Cross-Cultural Risk Factor Questionnaire, which assesses retrospectively a range of sociocultural and psychological ED risk factors occurring before the age of 12 years (46 predictors in total), was used. Results All three statistical approaches had satisfactory model accuracy, with an average area under the curve (AUC) of 86% for predicting ED onset and 70% for predicting AN v. BN. Predictive performance was greatest for the two regression methods (LR and LASSO), although the PRE technique relied on fewer predictors with comparable accuracy. The individual risk factors differed depending on the outcome classification (EDs v. non-EDs and AN v. BN). Conclusions Even though the conventional LR performed comparably to the ML approaches in terms of predictive accuracy, the ML methods produced more parsimonious predictive models. ML approaches offer a viable way to modify screening practices for ED risk that balance accuracy against participant burden.

KW - Anorexia nervosa

KW - bulimia nervosa

KW - eating disorders

KW - machine learning

KW - risk and protective factors

UR - http://www.scopus.com/inward/record.url?scp=85120628148&partnerID=8YFLogxK

U2 - 10.1017/S003329172100489X

DO - 10.1017/S003329172100489X

M3 - Article

AN - SCOPUS:85120628148

SN - 0033-2917

SP - 1

EP - 10

JO - Psychological Medicine

JF - Psychological Medicine

ER -