Applications of machine learning algorithms to predict therapeutic outcomes in depression: A meta-analysis and systematic review.

BACKGROUND No previous study has comprehensively reviewed the application of machine learning algorithms in mood disorders populations. Herein, we qualitatively and quantitatively evaluate previous studies of machine learning-devised models that predict therapeutic outcomes in mood disorders populations. METHODS We searched Ovid MEDLINE/PubMed from inception to February 8, 2018 for relevant studies that included adults with bipolar or unipolar depression; assessed therapeutic outcomes with a pharmacological, neuromodulatory, or manual-based psychotherapeutic intervention for depression; applied a machine learning algorithm; and reported predictors of therapeutic response. A random-effects meta-analysis of proportions and meta-regression analyses were conducted. RESULTS We identified 639 records: 75 full-text publications were assessed for eligibility; 26 studies (n=17,499) and 20 studies (n=6325) were included in qualitative and quantitative review, respectively. Classification algorithms were able to predict therapeutic outcomes with an overall accuracy of 0.82 (95% confidence interval [CI] of [0.77, 0.87]). Pooled estimates of classification accuracy were significantly greater (p < 0.01) in models informed by multiple data types (e.g., composite of phenomenological patient features and neuroimaging or peripheral gene expression data; pooled proportion [95% CI] = 0.93[0.86, 0.97]) when compared to models with lower-dimension data types (pooledproportion=0.68[0.62,0.74]to0.85[0.81,0.88]). LIMITATIONS Most studies were retrospective; differences in machine learning algorithms and their implementation (e.g., cross-validation, hyperparameter tuning); cannot infer importance of individual variables fed into learning algorithm. CONCLUSIONS Machine learning algorithms provide a powerful conceptual and analytic framework capable of integrating multiple data types and sources. An integrative approach may more effectively model neurobiological components as functional modules of pathophysiology embedded within the complex, social dynamics that influence the phenomenology of mental disorders.

[1]  R. Bagby,et al.  LATENT CLASSES OF NONRESPONDERS, RAPID RESPONDERS, AND GRADUAL RESPONDERS IN DEPRESSED OUTPATIENTS RECEIVING ANTIDEPRESSANT MEDICATION AND PSYCHOTHERAPY , 2015, Depression and anxiety.

[2]  Predisposition to treatment response in major depressive episode: A peripheral blood gene coexpression network analysis. , 2016, Journal of psychiatric research.

[3]  C. Begg,et al.  Operating characteristics of a rank correlation test for publication bias. , 1994, Biometrics.

[4]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[5]  Tianxi Cai,et al.  Testing a machine-learning algorithm to predict the persistence and severity of major depressive disorder from baseline self-reports , 2015, Molecular Psychiatry.

[6]  Theo Stijnen,et al.  The binomial distribution of meta-analysis was preferred to model within-study variability. , 2008, Journal of clinical epidemiology.

[7]  B. Lebowitz,et al.  Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. , 2006, The American journal of psychiatry.

[8]  J. Price,et al.  Machine learning approaches for integrating clinical and imaging features in late‐life depression classification and response prediction , 2015, International journal of geriatric psychiatry.

[9]  M. Rietschel,et al.  Combining clinical variables to optimize prediction of antidepressant treatment outcomes. , 2016, Journal of psychiatric research.

[10]  M. Trivedi,et al.  Pro-Inflammatory Cytokines as Predictors of Antidepressant Effects of Exercise in Major Depressive Disorder , 2012, Molecular Psychiatry.

[11]  M. Frank,et al.  Computational psychiatry as a bridge from neuroscience to clinical applications , 2016, Nature Neuroscience.

[12]  Kim-Anh Lê Cao,et al.  mixOmics: An R package for ‘omics feature selection and multiple data integration , 2017, bioRxiv.

[13]  A. Meyer-Lindenberg,et al.  Machine Learning for Precision Psychiatry: Opportunities and Challenges. , 2017, Biological psychiatry. Cognitive neuroscience and neuroimaging.

[14]  H. Scholte,et al.  A functional MRI marker may predict the outcome of electroconvulsive therapy in severe and treatment-resistant depression , 2014, Molecular Psychiatry.

[15]  R. Lam,et al.  Efficacy of Vortioxetine on Cognitive Functioning in Working Patients With Major Depressive Disorder. , 2017, The Journal of clinical psychiatry.

[16]  A. Danese,et al.  Inflammation and clinical response to treatment in depression: A meta-analysis , 2015, European Neuropsychopharmacology.

[17]  F. Jain,et al.  PREDICTIVE SOCIOECONOMIC AND CLINICAL PROFILES OF ANTIDEPRESSANT RESPONSE AND REMISSION , 2013, Depression and anxiety.

[18]  Shantanu H. Joshi,et al.  Methylphenidate modifies the motion of the circadian clock Lamotrigine in mood disorders and cocaine dependence Cortical glutamate in postpartum depression Effect of Electroconvulsive Therapy on Striatal Morphometry in Major Depressive Disorder , 2016 .

[19]  Leanne M Williams,et al.  A Cognitive–Emotional Biomarker for Predicting Remission with Antidepressant Medications: A Report from the iSPOT-D Trial , 2015, Neuropsychopharmacology.

[20]  Wolfgang Viechtbauer,et al.  Conducting Meta-Analyses in R with the metafor Package , 2010 .

[21]  Karl J. Friston,et al.  Computational Nosology and Precision Psychiatry , 2017, Computational Psychiatry.

[22]  M. Aerts,et al.  Metaprop: a Stata command to perform meta-analysis of binomial data , 2014, Archives of Public Health.

[23]  M. Maes,et al.  Peripheral Alterations in Cytokine and Chemokine Levels After Antidepressant Drug Treatment for Major Depressive Disorder: Systematic Review and Meta-Analysis , 2017, Molecular Neurobiology.

[24]  R. Atun,et al.  Estimating the true global burden of mental illness. , 2016, The lancet. Psychiatry.

[25]  Aamir Saeed Malik,et al.  A wavelet-based technique to predict treatment outcome for Major Depressive Disorder , 2017, PloS one.

[26]  Jan N. van Rijn,et al.  An Empirical Study of Hyperparameter Importance Across Datasets , 2017, AutoML@PKDD/ECML.

[27]  G. Tseng,et al.  Testing the Predictive Value of Peripheral Gene Expression for Nonremission Following Citalopram Treatment for Major Depression , 2015, Neuropsychopharmacology.

[28]  E. Gordon,et al.  Magnetic Resonance Imaging Measures of Brain Structure to Predict Antidepressant Treatment Outcome in Major Depressive Disorder , 2014, EBioMedicine.

[29]  Roger S McIntyre,et al.  Does Pharmacogenomic Testing Improve Clinical Outcomes for Major Depressive Disorder? A Systematic Review of Clinical Trials and Cost-Effectiveness Studies. , 2017, The Journal of clinical psychiatry.

[30]  R. Tibshirani,et al.  Classification and prediction of clinical Alzheimer's diagnosis based on plasma signaling proteins , 2007, Nature Medicine.

[31]  R. Tweedie,et al.  A Nonparametric “Trim and Fill” Method of Accounting for Publication Bias in Meta-Analysis , 2000 .

[32]  D. Parker,et al.  Guidance for conducting systematic scoping reviews , 2015, International journal of evidence-based healthcare.

[33]  A. McIntosh.  Toward a Systems Biology of Mood Disorder , 2013, Biological Psychiatry.

[34]  Pia Baldinger,et al.  The combined effect of genetic polymorphisms and clinical parameters on treatment outcome in treatment-resistant depression , 2015, European Neuropsychopharmacology.

[35]  Harlan M. Krumholz,et al.  Reevaluating the Efficacy and Predictability of Antidepressant Treatments: A Symptom Clustering Approach , 2017, JAMA psychiatry.

[36]  Kenneth S. Kendler,et al.  Toward a comprehensive developmental model for major depression in women. , 2002 .

[37]  R. McIntyre,et al.  Brain-derived neurotrophic factor, impaired glucose metabolism, and bipolar disorder course. , 2016, Bipolar disorders.

[38]  R. McIntyre,et al.  Cognition in major depressive disorder: a ‘Systemically Important Functional Index’ (SIFI) , 2016, Current opinion in psychiatry.

[39]  S. Cole,et al.  Clustering of Depression and Inflammation in Adolescents Previously Exposed to Childhood Adversity , 2012, Biological Psychiatry.

[40]  Andrew T. Drysdale,et al.  Resting-state connectivity biomarkers define neurophysiological subtypes of depression , 2016, Nature Medicine.

[41]  A. Rush,et al.  Increase in work productivity of depressed individuals with improvement in depressive symptom severity. , 2013, The American journal of psychiatry.

[42]  G. Molenberghs,et al.  An extended random-effects approach to modeling repeated, overdispersed count data , 2007, Lifetime data analysis.

[43]  Rupert Lanzenberger,et al.  Refining Prediction in Treatment-Resistant Depression: Results of Machine Learning Analyses in the TRD III Sample. , 2017, The Journal of clinical psychiatry.

[44]  Ben D. Fulcher,et al.  Responders to rTMS for depression show increased fronto-midline theta and theta connectivity compared to non-responders , 2018, Brain Stimulation.

[45]  V. Arolt,et al.  Prediction of Individual Response to Electroconvulsive Therapy via Machine Learning on Structural Magnetic Resonance Imaging Data. , 2016, JAMA psychiatry.

[46]  A. Mechelli,et al.  Using Support Vector Machine to identify imaging biomarkers of neurological and psychiatric disease: A critical review , 2012, Neuroscience & Biobehavioral Reviews.

[47]  M. Trivedi,et al.  Symptom clusters as predictors of late response to antidepressant treatment. , 2005, The Journal of clinical psychiatry.

[48]  Donel M. Martin,et al.  Predicting tDCS treatment outcomes of patients with major depressive disorder using automated EEG classification. , 2017, Journal of affective disorders.

[49]  Enrico Smeraldi,et al.  A neural network model for combining clinical predictors of antidepressant response in mood disorders. , 2007, Journal of affective disorders.

[50]  D. F. Drake,et al.  A randomized controlled trial of the tumor necrosis factor antagonist infliximab for treatment-resistant depression: the role of baseline inflammatory biomarkers. , 2013, JAMA psychiatry.

[51]  E. S. Pearson,et al.  THE USE OF CONFIDENCE OR FIDUCIAL LIMITS ILLUSTRATED IN THE CASE OF THE BINOMIAL , 1934 .

[52]  D. Moher,et al.  Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. , 2009, Journal of clinical epidemiology.

[53]  Enrico Smeraldi,et al.  Neural network analysis in pharmacogenetics of mood disorders , 2004, BMC Medical Genetics.

[54]  M. Ghassemi,et al.  Predicting early psychiatric readmission with natural language processing of narrative discharge summaries , 2016, Translational psychiatry.

[55]  Marcia K. Johnson,et al.  Cross-trial prediction of treatment outcome in depression: a machine learning approach. , 2016, The lancet. Psychiatry.

[56]  S Duval,et al.  Trim and Fill: A Simple Funnel‐Plot–Based Method of Testing and Adjusting for Publication Bias in Meta‐Analysis , 2000, Biometrics.

[57]  G. Rakesh.  Suicide Prediction With Machine Learning , 2017 .

[58]  M. Brammer,et al.  Neuroanatomy of verbal working memory as a diagnostic biomarker for depression , 2008, Neuroreport.

[59]  Janaina Mourao-Miranda,et al.  Neural correlates of sad faces predict clinical remission to cognitive behavioural therapy in depression , 2009, Neuroreport.

[60]  Ahmad Khodayari-Rostamabad,et al.  Using pre-treatment electroencephalography data to predict response to transcranial magnetic stimulation therapy for major depression , 2011, 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[61]  T. Insel,et al.  Brain disorders? Precisely , 2015, Science.

[62]  J. Tukey,et al.  Transformations Related to the Angular and the Square Root , 1950 .

[63]  Huafu Chen,et al.  Classification of Different Therapeutic Responses of Major Depressive Disorder with Multivariate Pattern Analysis Method Based on Structural MR Scans , 2012, PloS one.

[64]  Ahmad Khodayari-Rostamabad,et al.  A machine learning approach using EEG data to predict response to SSRI treatment for major depressive disorder , 2013, Clinical Neurophysiology.

[65]  M. Trivedi,et al.  Peripheral biomarkers of major depression and antidepressant treatment response: Current knowledge and future outlooks. , 2017, Journal of affective disorders.

[66]  A. Barabasi,et al.  Network medicine--from obesity to the "diseasome". , 2007, The New England journal of medicine.

[67]  R. Cameron Craddock,et al.  Data-Driven Phenotypic Categorization for Neurobiological Analyses: Beyond DSM-5 Labels , 2016, Biological Psychiatry.