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

Lee, Yena; Ragguett, Renee-Marie; Mansur, Rodrigo B.; Boutilier, Justin J.; Rosenblat, Joshua D.; Trevizol, Alisson Paulino; Brietzke, Elisa; Lin, Kangguang; Pan, Zihang; Subramaniapillai, Mehala; Chan, Timothy C. Y.; Fus, Dominika; Park, Caroline; Musial, Natalie; Zuckerman, Hannah; Chen, Vincent Chin‐Hung; Ho, Roger; Rong, Carola; McIntyre, Roger S.

doi:10.1016/j.jad.2018.08.073

Cited by 330 publications

(168 citation statements)

References 62 publications

Supporting

Mentioning

154

Contrasting

Unclassified

Order By: Relevance

“…This examination was not able to take a multivariable approach to predicting treatment-response, which will likely provide more valid and replicable predictions through integrating diverse information (e.g. inflammatory and clinical variables) rather than considering singular factors (Lee et al, 2018).…”

Section: Recommendations and Clinical Implicationsmentioning

confidence: 99%

Inflammatory profiles of severe treatment-resistant depression

Strawbridge

Hodsoll

Powell

et al. 2019

Journal of Affective Disorders

View full text Add to dashboard Cite

show abstract

Section: Recommendations and Clinical Implicationsmentioning

confidence: 99%

Inflammatory profiles of severe treatment-resistant depression

Strawbridge

Hodsoll

Powell

et al. 2019

Journal of Affective Disorders

View full text Add to dashboard Cite

show abstract

“…Similarly, adverse social and environmental influences (e.g., early childhood adversity, sexual abuse, socioeconomic status, and social isolation) are well recognized to increase the risk for mood disorder incidence and moderate illness course and trajectory. However, independently, these factors do not consistently predict response and remission …”

Section: Interindividual Variability and Personalized Treatment Selecmentioning

confidence: 97%

“…However, independently, these factors do not consistently predict response and remission. 85 In order to address the underlying factors that influence response, EEG studies have been performed. Changes in the resting-state quantitative EEG and theta cordance over the first week of HFL over the left DLPFC were investigated as biomarkers for response prediction at the end of 6 weeks of treatment (N = 18).…”

Section: Interindividual Variability and Personalized Treatment Selecmentioning

confidence: 99%

An Update on Repetitive Transcranial Magnetic Stimulation for the Treatment of Major Depressive Disorder

Trevizol

Blumberger

2019

Clin Pharma and Therapeutics

View full text Add to dashboard Cite

Repetitive transcranial magnetic stimulation (rTMS) has emerged as an evidenced‐based treatment for major depression that does not respond to standard first‐line therapies. The majority of data support the use of high‐frequency (10 Hz) treatment delivered to the left dorsolateral prefrontal cortex. Intermittent theta burst stimulation is a new emerging treatment that reduces the time required to deliver treatment and can increase capacity and access to this treatment. This review will comprehensively cover recent advancements in the field of rTMS for depression, including stimulation parameters and targets aimed at enhancing outcomes. In addition, efforts to use modern neuroscience tools to personalize this treatment and optimize outcomes will be reviewed.

show abstract

“…In contrast, machine learning allows for individualized prediction through the implementation of learning algorithms, which make fewer assumptions about data-generation, to find patterns in large, heterogeneous datasets. Advances in machine learning have highlighted its utility in identifying patterns in complex data for psychiatric research (Iniesta et al, 2016;Passos et al, 2016) and specifically for outcomes of depression treatments (Lee et al, 2018). Recent studies have leveraged machine learning methods to predict antidepressant treatment response for individuals with depression, identifying 25 features most predictive of whether a patient will respond to citalopram (Chekroud et al, 2016), predicting persistence, chronicity, and severity of depression from self-report questionnaires (Kessler et al, 2016), predicting treatment response to electroconvulsive therapy (ECT) using baseline hippocampal subfield volumes (Cao et al, 2018), predicting treatment resistance before initiation of a second antidepressant (Nie et al, 2018), using deep learning to predict response to SSRIs (Lin et al, 2018), and using Random Forests to predict outcome in treatmentresistant depression (Kautzky et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

Analysis of Features Selected by a Deep Learning Model for Differential Treatment Selection in Depression

Mehltretter

Rollins

Benrimoh

et al. 2020

Front. Artif. Intell.

View full text Add to dashboard Cite

Background: Deep learning has utility in predicting differential antidepressant treatment response among patients with major depressive disorder, yet there remains a paucity of research describing how to interpret deep learning models in a clinically or etiologically meaningful way. In this paper, we describe methods for analyzing deep learning models of clinical and demographic psychiatric data, using our recent work on a deep learning model of STAR * D and CO-MED remission prediction. Methods: Our deep learning analysis with STAR * D and CO-MED yielded four models that predicted response to the four treatments used across the two datasets. Here, we use classical statistics and simple data representations to improve interpretability of the features output by our deep learning model and provide finer grained understanding of their clinical and etiological significance. Specifically, we use representations derived from our model to yield features predicting both treatment non-response and differential treatment response to four standard antidepressants, and use linear regression and t-tests to address questions about the contribution of trauma, education, and somatic symptoms to our models. Results: Traditional statistics were able to probe the input features of our deep learning models, reproducing results from previous research, while providing novel insights into depression causes and treatments. We found that specific features were predictive of treatment response, and were able to break these down by treatment and non-response categories; that specific trauma indices were differentially predictive of baseline depression severity; that somatic symptoms were significantly different between males and females, and that education and low income proved important psycho-social stressors associated with depression. Conclusion: Traditional statistics can augment interpretation of deep learning models. Such interpretation can lend us new hypotheses about depression and contribute to building causal models of etiology and prognosis. We discuss dataset-specific effects and ideal clinical samples for machine learning analysis aimed at improving tools to assist in optimizing treatment.

show abstract

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

Cited by 330 publications

References 62 publications

Inflammatory profiles of severe treatment-resistant depression

Inflammatory profiles of severe treatment-resistant depression

An Update on Repetitive Transcranial Magnetic Stimulation for the Treatment of Major Depressive Disorder

Analysis of Features Selected by a Deep Learning Model for Differential Treatment Selection in Depression

Contact Info

Product

Resources

About