Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder

Zuo Zhang; Lauren Robinson; Robert Whelan; Lee Jollans; Zijian Wang; Frauke Nees; Congying Chu; Marina Bobou; Dongping Du; Ilinca Cristea; Tobias Banaschewski; Gareth J. Barker; Arun L.W. Bokde; Antoine Grigis; Hugh Garavan; Andreas Heinz; Rüdiger Brühl; Jean-Luc Martinot; Marie-Laure Paillère Martinot; Eric Artiges; Dimitri Papadopoulos Orfanos; Luise Poustka; Sarah Hohmann; Sabina Millenet; Juliane H. Fröhner; Michael N. Smolka; Nilakshi Vaidya; Henrik Walter; Jeanne Winterer; M. John Broulidakis; Betteke Maria van Noort; Argyris Stringaris; Jani Penttilä; Yvonne Grimmer; Corinna Insensee; Andreas Becker; Yuning Zhang; Sinead King; Julia Sinclair; Gunter Schumann; Ulrike Schmidt; Sylvane Desrivières

doi:10.1016/j.jad.2024.12.053

Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder

Zuo Zhang, Lauren Robinson, Robert Whelan, Lee Jollans, Zijian Wang, Frauke Nees, Congying Chu, Marina Bobou, Dongping Du, Ilinca Cristea, Tobias Banaschewski, Gareth J. Barker, Arun L.W. Bokde, Antoine Grigis, Hugh Garavan, Andreas Heinz, Rüdiger Brühl, Jean-Luc Martinot, Marie-Laure Paillère Martinot, Eric ArtigesDimitri Papadopoulos Orfanos, Luise Poustka, Sarah Hohmann, Sabina Millenet, Juliane H. Fröhner, Michael N. Smolka, Nilakshi Vaidya, Henrik Walter, Jeanne Winterer, M. John Broulidakis, Betteke Maria van Noort, Argyris Stringaris, Jani Penttilä, Yvonne Grimmer, Corinna Insensee, Andreas Becker, Yuning Zhang, Sinead King, Julia Sinclair, Gunter Schumann, Ulrike Schmidt, Sylvane Desrivières

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

223 Downloads (Pure)

Abstract

Background
Early diagnosis and treatment of mental illnesses is hampered by the lack of reliable markers. This study used machine learning models to uncover diagnostic and risk prediction markers for eating disorders (EDs), major depressive disorder (MDD), and alcohol use disorder (AUD).
Methods
Case-control samples (aged 18–25 years), including participants with Anorexia Nervosa (AN), Bulimia Nervosa (BN), MDD, AUD, and matched controls, were used for diagnostic classification. For risk prediction, we used a longitudinal population-based sample (IMAGEN study), assessing adolescents at ages 14, 16 and 19. Regularized logistic regression models incorporated broad data domains spanning psychopathology, personality, cognition, substance use, and environment.
Results
The classification of EDs was highly accurate, even when excluding body mass index from the analysis. The area under the receiver operating characteristic curves (AUC-ROC [95 % CI]) reached 0.92 [0.86–0.97] for AN and 0.91 [0.85–0.96] for BN. The classification accuracies for MDD (0.91 [0.88–0.94]) and AUD (0.80 [0.74–0.85]) were also high. The models demonstrated high transdiagnostic potential, as those trained for EDs were also accurate in classifying AUD and MDD from healthy controls, and vice versa (AUC-ROCs, 0.75–0.93). Shared predictors, such as neuroticism, hopelessness, and symptoms of attention-deficit/hyperactivity disorder, were identified as reliable classifiers. In the longitudinal population sample, the models exhibited moderate performance in predicting the development of future ED symptoms (0.71 [0.67–0.75]), depressive symptoms (0.64 [0.60–0.68]), and harmful drinking (0.67 [0.64–0.70]).
Conclusions
Our findings demonstrate the potential of combining multi-domain data for precise diagnostic and risk prediction applications in psychiatry.

Original language	English
Journal	Journal of affective disorders
Early online date	17 Dec 2024
DOIs	https://doi.org/10.1016/j.jad.2024.12.053
Publication status	E-pub ahead of print - 17 Dec 2024

UN SDGs

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

Access to Document

10.1016/j.jad.2024.12.053

eb35ffa6-6126-42d2-9a52-ac076731c268
This work is licensed under a CC BY 4.0 License
Submitted manuscript, 1.02 MBLicence: CC BY
Machine learning models for_ZHANG_Publishedonline17December2024_GREEN_AAMAccepted author manuscript, 1.58 MB

1 Article

Distinct personality profiles associated with disease risk and diagnostic status in eating disorders
IMAGEN, ESTRA & STRATIFY, 31 May 2024, In: Journal of Affective Disorders. 360, p. 146-155 10 p.
Research output: Contribution to journal › Article › peer-review

Open Access
2 Citations (Scopus)

Cite this

Zhang, Z., Robinson, L., Whelan, R., Jollans, L., Wang, Z., Nees, F., Chu, C., Bobou, M., Du, D., Cristea, I., Banaschewski, T., Barker, G. J., Bokde, A. L. W., Grigis, A., Garavan, H., Heinz, A., Brühl, R., Martinot, J.-L., Martinot, M.-L. P., ... Desrivières, S. (2024). Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder. Journal of affective disorders. Advance online publication. https://doi.org/10.1016/j.jad.2024.12.053

@article{a9363143630f48afa3b0dd7618d69384,

title = "Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder",

abstract = "BackgroundEarly diagnosis and treatment of mental illnesses is hampered by the lack of reliable markers. This study used machine learning models to uncover diagnostic and risk prediction markers for eating disorders (EDs), major depressive disorder (MDD), and alcohol use disorder (AUD).MethodsCase-control samples (aged 18–25 years), including participants with Anorexia Nervosa (AN), Bulimia Nervosa (BN), MDD, AUD, and matched controls, were used for diagnostic classification. For risk prediction, we used a longitudinal population-based sample (IMAGEN study), assessing adolescents at ages 14, 16 and 19. Regularized logistic regression models incorporated broad data domains spanning psychopathology, personality, cognition, substance use, and environment.ResultsThe classification of EDs was highly accurate, even when excluding body mass index from the analysis. The area under the receiver operating characteristic curves (AUC-ROC [95 % CI]) reached 0.92 [0.86–0.97] for AN and 0.91 [0.85–0.96] for BN. The classification accuracies for MDD (0.91 [0.88–0.94]) and AUD (0.80 [0.74–0.85]) were also high. The models demonstrated high transdiagnostic potential, as those trained for EDs were also accurate in classifying AUD and MDD from healthy controls, and vice versa (AUC-ROCs, 0.75–0.93). Shared predictors, such as neuroticism, hopelessness, and symptoms of attention-deficit/hyperactivity disorder, were identified as reliable classifiers. In the longitudinal population sample, the models exhibited moderate performance in predicting the development of future ED symptoms (0.71 [0.67–0.75]), depressive symptoms (0.64 [0.60–0.68]), and harmful drinking (0.67 [0.64–0.70]).ConclusionsOur findings demonstrate the potential of combining multi-domain data for precise diagnostic and risk prediction applications in psychiatry.",

author = "Zuo Zhang and Lauren Robinson and Robert Whelan and Lee Jollans and Zijian Wang and Frauke Nees and Congying Chu and Marina Bobou and Dongping Du and Ilinca Cristea and Tobias Banaschewski and Barker, {Gareth J.} and Bokde, {Arun L.W.} and Antoine Grigis and Hugh Garavan and Andreas Heinz and R{\"u}diger Br{\"u}hl and Jean-Luc Martinot and Martinot, {Marie-Laure Paill{\`e}re} and Eric Artiges and Orfanos, {Dimitri Papadopoulos} and Luise Poustka and Sarah Hohmann and Sabina Millenet and Fr{\"o}hner, {Juliane H.} and Smolka, {Michael N.} and Nilakshi Vaidya and Henrik Walter and Jeanne Winterer and Broulidakis, {M. John} and {van Noort}, {Betteke Maria} and Argyris Stringaris and Jani Penttil{\"a} and Yvonne Grimmer and Corinna Insensee and Andreas Becker and Yuning Zhang and Sinead King and Julia Sinclair and Gunter Schumann and Ulrike Schmidt and Sylvane Desrivi{\`e}res",

year = "2024",

month = dec,

day = "17",

doi = "10.1016/j.jad.2024.12.053",

language = "English",

journal = "Journal of affective disorders",

issn = "0165-0327",

publisher = "Elsevier",

}

Zhang, Z , Robinson, L, Whelan, R, Jollans, L, Wang, Z, Nees, F, Chu, C , Bobou, M, Du, D, Cristea, I, Banaschewski, T, Barker, GJ, Bokde, ALW, Grigis, A, Garavan, H, Heinz, A, Brühl, R, Martinot, J-L, Martinot, M-LP, Artiges, E, Orfanos, DP, Poustka, L, Hohmann, S, Millenet, S, Fröhner, JH, Smolka, MN, Vaidya, N, Walter, H, Winterer, J, Broulidakis, MJ, van Noort, BM, Stringaris, A, Penttilä, J, Grimmer, Y, Insensee, C, Becker, A, Zhang, Y , King, S, Sinclair, J, Schumann, G , Schmidt, U & Desrivières, S 2024, 'Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder', Journal of affective disorders. https://doi.org/10.1016/j.jad.2024.12.053

TY - JOUR

T1 - Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder

AU - Zhang, Zuo

AU - Robinson, Lauren

AU - Whelan, Robert

AU - Jollans, Lee

AU - Wang, Zijian

AU - Nees, Frauke

AU - Chu, Congying

AU - Bobou, Marina

AU - Du, Dongping

AU - Cristea, Ilinca

AU - Banaschewski, Tobias

AU - Barker, Gareth J.

AU - Bokde, Arun L.W.

AU - Grigis, Antoine

AU - Garavan, Hugh

AU - Heinz, Andreas

AU - Brühl, Rüdiger

AU - Martinot, Jean-Luc

AU - Martinot, Marie-Laure Paillère

AU - Artiges, Eric

AU - Orfanos, Dimitri Papadopoulos

AU - Poustka, Luise

AU - Hohmann, Sarah

AU - Millenet, Sabina

AU - Fröhner, Juliane H.

AU - Smolka, Michael N.

AU - Vaidya, Nilakshi

AU - Walter, Henrik

AU - Winterer, Jeanne

AU - Broulidakis, M. John

AU - van Noort, Betteke Maria

AU - Stringaris, Argyris

AU - Penttilä, Jani

AU - Grimmer, Yvonne

AU - Insensee, Corinna

AU - Becker, Andreas

AU - Zhang, Yuning

AU - King, Sinead

AU - Sinclair, Julia

AU - Schumann, Gunter

AU - Schmidt, Ulrike

AU - Desrivières, Sylvane

PY - 2024/12/17

Y1 - 2024/12/17

N2 - BackgroundEarly diagnosis and treatment of mental illnesses is hampered by the lack of reliable markers. This study used machine learning models to uncover diagnostic and risk prediction markers for eating disorders (EDs), major depressive disorder (MDD), and alcohol use disorder (AUD).MethodsCase-control samples (aged 18–25 years), including participants with Anorexia Nervosa (AN), Bulimia Nervosa (BN), MDD, AUD, and matched controls, were used for diagnostic classification. For risk prediction, we used a longitudinal population-based sample (IMAGEN study), assessing adolescents at ages 14, 16 and 19. Regularized logistic regression models incorporated broad data domains spanning psychopathology, personality, cognition, substance use, and environment.ResultsThe classification of EDs was highly accurate, even when excluding body mass index from the analysis. The area under the receiver operating characteristic curves (AUC-ROC [95 % CI]) reached 0.92 [0.86–0.97] for AN and 0.91 [0.85–0.96] for BN. The classification accuracies for MDD (0.91 [0.88–0.94]) and AUD (0.80 [0.74–0.85]) were also high. The models demonstrated high transdiagnostic potential, as those trained for EDs were also accurate in classifying AUD and MDD from healthy controls, and vice versa (AUC-ROCs, 0.75–0.93). Shared predictors, such as neuroticism, hopelessness, and symptoms of attention-deficit/hyperactivity disorder, were identified as reliable classifiers. In the longitudinal population sample, the models exhibited moderate performance in predicting the development of future ED symptoms (0.71 [0.67–0.75]), depressive symptoms (0.64 [0.60–0.68]), and harmful drinking (0.67 [0.64–0.70]).ConclusionsOur findings demonstrate the potential of combining multi-domain data for precise diagnostic and risk prediction applications in psychiatry.

AB - BackgroundEarly diagnosis and treatment of mental illnesses is hampered by the lack of reliable markers. This study used machine learning models to uncover diagnostic and risk prediction markers for eating disorders (EDs), major depressive disorder (MDD), and alcohol use disorder (AUD).MethodsCase-control samples (aged 18–25 years), including participants with Anorexia Nervosa (AN), Bulimia Nervosa (BN), MDD, AUD, and matched controls, were used for diagnostic classification. For risk prediction, we used a longitudinal population-based sample (IMAGEN study), assessing adolescents at ages 14, 16 and 19. Regularized logistic regression models incorporated broad data domains spanning psychopathology, personality, cognition, substance use, and environment.ResultsThe classification of EDs was highly accurate, even when excluding body mass index from the analysis. The area under the receiver operating characteristic curves (AUC-ROC [95 % CI]) reached 0.92 [0.86–0.97] for AN and 0.91 [0.85–0.96] for BN. The classification accuracies for MDD (0.91 [0.88–0.94]) and AUD (0.80 [0.74–0.85]) were also high. The models demonstrated high transdiagnostic potential, as those trained for EDs were also accurate in classifying AUD and MDD from healthy controls, and vice versa (AUC-ROCs, 0.75–0.93). Shared predictors, such as neuroticism, hopelessness, and symptoms of attention-deficit/hyperactivity disorder, were identified as reliable classifiers. In the longitudinal population sample, the models exhibited moderate performance in predicting the development of future ED symptoms (0.71 [0.67–0.75]), depressive symptoms (0.64 [0.60–0.68]), and harmful drinking (0.67 [0.64–0.70]).ConclusionsOur findings demonstrate the potential of combining multi-domain data for precise diagnostic and risk prediction applications in psychiatry.

U2 - 10.1016/j.jad.2024.12.053

DO - 10.1016/j.jad.2024.12.053

M3 - Article

C2 - 38352452

SN - 0165-0327

JO - Journal of affective disorders

JF - Journal of affective disorders

ER -

Machine learning models for diagnosis and risk prediction in eating disorders, depression, and alcohol use disorder

Abstract

UN SDGs

Access to Document

Fingerprint

Research output

Distinct personality profiles associated with disease risk and diagnostic status in eating disorders

Cite this