Multi-site benchmark classification of major depressive disorder using machine learning on cortical and subcortical measures
- PMID: 38212349
- PMCID: PMC10784593
- DOI: 10.1038/s41598-023-47934-8
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 = 5365) to provide a generalizable ML classification benchmark of major depressive disorder (MDD) using shallow linear and non-linear models. Leveraging brain measures from standardized ENIGMA analysis pipelines in FreeSurfer, we were able to classify MDD versus healthy controls (HC) with a balanced accuracy of around 62%. But after harmonizing the data, e.g., using ComBat, the balanced accuracy dropped to approximately 52%. Accuracy results close to random chance levels were also 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 yield more encouraging prospects.
© 2024. The Author(s).
Conflict of interest statement
PMT and NJ received a research grant from Biogen, Inc., for research unrelated to this manuscript. HJG has received travel grants and speakers honoraria from Fresenius Medical Care, Neuraxpharm, Servier and Janssen Cilag as well as research funding from Fresenius Medical Care unrelated to this manuscript. JCS has served as a consultant for Pfizer,Sunovion, Sanofi, Johnson & Johnson, Livanova, and Boehringer Ingelheim. The remaining authors declare no conflict of interest.
Figures
Similar articles
-
Graph convolutional network with attention mechanism improve major depressive depression diagnosis based on plasma biomarkers and neuroimaging data.J Affect Disord. 2024 Sep 1;360:336-344. doi: 10.1016/j.jad.2024.05.136. Epub 2024 May 31. J Affect Disord. 2024. PMID: 38824965
-
A systematic review on the potential use of machine learning to classify major depressive disorder from healthy controls using resting state fMRI measures.Neurosci Biobehav Rev. 2023 Jan;144:104972. doi: 10.1016/j.neubiorev.2022.104972. Epub 2022 Nov 24. Neurosci Biobehav Rev. 2023. PMID: 36436736 Review.
-
The alterations of brain functional connectivity networks in major depressive disorder detected by machine learning through multisite rs-fMRI data.Behav Brain Res. 2022 Oct 28;435:114058. doi: 10.1016/j.bbr.2022.114058. Epub 2022 Aug 20. Behav Brain Res. 2022. PMID: 35995263
-
Will machine learning applied to neuroimaging in bipolar disorder help the clinician? A critical review and methodological suggestions.Bipolar Disord. 2020 Jun;22(4):334-355. doi: 10.1111/bdi.12895. Epub 2020 Mar 20. Bipolar Disord. 2020. PMID: 32108409 Review.
-
Subcortical brain structure and suicidal behaviour in major depressive disorder: a meta-analysis from the ENIGMA-MDD working group.Transl Psychiatry. 2017 May 2;7(5):e1116. doi: 10.1038/tp.2017.84. Transl Psychiatry. 2017. PMID: 28463239 Free PMC article.
Cited by
-
Systematic Review: White Matter Microstructural Organization in Adolescents With Depression.JAACAP Open. 2023 Dec;1(4):233-245. doi: 10.1016/j.jaacop.2023.08.006. Epub 2023 Sep 5. JAACAP Open. 2023. PMID: 38576601 Free PMC article.
References
-
- Cleare AFSWCDKMBMLPAJ. A multidimensional tool to quantify treatment resistance in depression: The Maudsley staging method. J. Clin. Psychiatry. 2009;70:12363. - PubMed
MeSH terms
Grants and funding
- R37 MH101495/MH/NIMH NIH HHS/United States
- P30 CA016672/CA/NCI NIH HHS/United States
- P41 RR008079/RR/NCRR NIH HHS/United States
- G0802594/MRC_/Medical Research Council/United Kingdom
- MR/L010305/1/MRC_/Medical Research Council/United Kingdom
- R01 MH116147/MH/NIMH NIH HHS/United States
- R01 HD050735/HD/NICHD NIH HHS/United States
- R01 CA193522/CA/NCI NIH HHS/United States
- K23 MH090421/MH/NIMH NIH HHS/United States
- R01 MH085734/MH/NIMH NIH HHS/United States
- U54 EB020403/EB/NIBIB NIH HHS/United States
- R01 MH117601/MH/NIMH NIH HHS/United States
- ZIA MH002857/ImNIH/Intramural NIH HHS/United States
- K01 MH117442/MH/NIMH NIH HHS/United States
- R01 NS073939/NS/NINDS NIH HHS/United States
LinkOut - more resources
Full Text Sources
Medical
Miscellaneous