Multi-site benchmark classification of major depressive disorder using machine learning on cortical and subcortical measures
Abstract
Machine learning (ML) techniques have gained popularity in the neuroimaging field due to their potential for classifying neuropsychiatric disorders. However, the diagnostic predictive power of the existing algorithms has been limited by small sample sizes, lack of representativeness, data leakage, and/or overfitting. Here, we overcome these limitations with the largest multi-site sample size to date (n=5,356) to provide a generalizable ML classification benchmark of major depressive disorder (MDD). Using brain measures from standardized ENIGMA analysis pipelines in FreeSurfer, we were able to classify MDD vs healthy controls (HC) with around 62% balanced accuracy, but when harmonizing the data using ComBat balanced accuracy dropped to approximately 52%. Similar results were observed in stratified groups according to age of onset, antidepressant use, number of episodes and sex. Future studies incorporating higher dimensional brain imaging/phenotype features, and/or using more advanced machine and deep learning methods may achieve more encouraging prospects.
Keywords
Cite
@article{arxiv.2206.08122,
title = {Multi-site benchmark classification of major depressive disorder using machine learning on cortical and subcortical measures},
author = {Vladimir Belov and Tracy Erwin-Grabner and Ali Saffet Gonul and Alyssa R. Amod and Amar Ojha and Andre Aleman and Annemiek Dols and Anouk Scharntee and Aslihan Uyar-Demir and Ben J Harrison and Benson M. Irungu and Bianca Besteher and Bonnie Klimes-Dougan and Brenda W. J. H. Penninx and Bryon A. Mueller and Carlos Zarate and Christopher G. Davey and Christopher R. K. Ching and Colm G. Connolly and Cynthia H. Y. Fu and Dan J. Stein and Danai Dima and David E. J. Linden and David M. A. Mehler and Edith Pomarol-Clotet and Elena Pozzi and Elisa Melloni and Francesco Benedetti and Frank P. MacMaster and Hans J. Grabe and Henry Völzke and Ian H. Gotlib and Jair C. Soares and Jennifer W. Evans and Kang Sim and Katharina Wittfeld and Kathryn Cullen and Liesbeth Reneman and Mardien L. Oudega and Margaret J. Wright and Maria J. Portella and Matthew D. Sacchet and Meng Li and Moji Aghajani and Mon-Ju Wu and Natalia Jaworska and Neda Jahanshad and Nic J. A. van der Wee and Nynke Groenewold and Paul J. Hamilton and Philipp Saemann and Robin Bülow and Sara Poletti and Sarah Whittle and Sophia I. Thomopoulos and Steven J. A. van and der Werff and Sheri-Michelle Koopowitz and Thomas Lancaster and Tiffany C. Ho and Tony T. Yang and Zeynep Basgoze and Dick J. Veltman and Lianne Schmaal and Paul M. Thompson and Roberto Goya-Maldonado},
journal= {arXiv preprint arXiv:2206.08122},
year = {2022}
}
Comments
main document 37 pages; supplementary material 24 pages