Using Machine Learning to Predict Remission in Patients with Major Depressive Disorder Treated with Desvenlafaxine

Abstract

Background: Major depressive disorder (MDD) is a common and burdensome condition that has low rates of treatment success. Although antidepressant medications are effective for MDD, remission rates are low and patients often require several medication switches to achieve remission. Hence, selecting an effective antidepressant is primarily determined by trial and error. Techniques using machine learning hold potential for predicting treatment success with a particular medication. This study uses baseline clinical data in creating machine learning models that learn to predict remission status after desvenlafaxine (DVS) treatment. Methods: We applied machine learning algorithms to data from 3776 MDD patients in nine phase-III/IV clinical trials, to produce a model predicting symptom remission, defined as an 8-week Hamilton Depression Rating Scale (HAM-D) score of 7. We trained the model on a randomly selected 90% of the data (n=3399), then evaluated that learned model on a holdout set (n=377). Outcomes: Our resulting classifier, a trained linear support vector machine (SVM), had a holdout set accuracy of 69·0%, significantly greater than the probability of classifying a patient correctly by chance. We demonstrate that this learning process is stable by repeatedly sampling part of the dataset and running the learner on this sample; these runs had an average accuracy of 67·0% +/- 1·8%. Our model, based on 26 clinical features, proved sufficient to predict DVS remission significantly better than chance. This may allow more accurate use of DVS without waiting 8 weeks to determine treatment outcome, and may serve as a first step towards changing psychiatric care by incorporating clinical assistive technologies using machine learned models. Funding: Data for this project were provided by Pfizer Inc. through a data sharing partnership with the University of Alberta. Declaration of Interest: SD, AG, RG, PC, and MB have no conflict of interest nor disclosure to make with regards to this paper. RL has received honoraria for ad hoc speaking or advising/consulting, or received research funds, from: Akili, Allergan, Asia-Pacific Economic Cooperation, BC Leading Edge Foundation, Canadian Institutes of Health Research, Canadian Network for Mood and Anxiety Treatments, Canadian Psychiatric Association, CME Institute, Hansoh, Healthy Minds Canada, Janssen, Lundbeck, Lundbeck Institute, Medscape, Mind.Me, Mitacs, Ontario Brain Institute, Otsuka, Pfizer, St. Jude Medical, University Health Network Foundation, and VGH-UBCH Foundation. BC is partially supported by the NARSAD Young Investigator award by the Brain & Behavior Research Foundation. JB received a studentship from Alberta Innovates: Health Solutions to support this work, and has previously received studentship/internship funding from the Natural Sciences and Engineering Research Council, Alberta Innovates: Technology Futures, and Mitacs. Ethical Approval: This study was approved by the University of Alberta Research Ethics Board, study Pro00064974, and all patients involved gave written consent for their anonymized data to be used.