Machine Learning Model Can Predict Treatment-Resistant Depression Outcomes
Applying a machine-learning algorithm can accurately predict outcomes in treatment-resistant major depressive disorder.
Applying a machine-learning algorithm can predict outcomes for treatment-resistant major depressive disorder with 75% accuracy, according to results published in the Journal of Clinical Psychiatry.
The study included patients diagnosed with major depressive disorder (MDD) according to DSM-IV criteria enrolled between 2011 and 2016. The researchers defined treatment-resistant depression (TRD) as not responding to antidepressant treatment, characterized by a Montgomery-Asberg Depression Rating Scale (MADRS) score below 22 after at least 2 antidepressant trials of adequate length and dosage.
The researchers used RandomForest (RF) to predict treatment outcome phenotypes in a 10-fold cross-validation. Using the full model with 47 predictors, the model's accuracy was 75.0%; when the number of predictors was reduced to 15, accuracies were between 67.6% and 71.0% for different test sets.
The most informative predictors of treatment outcome were baseline MADRS score for the current episode, impairment of family, social and work life, the timespan between first and last depressive episode, severity, suicidal risk, age, body mass index, and the number of lifetime depressive episodes, as well as lifetime duration of hospitalization.
Using these results, the researchers were able to develop a 15-predictor model that takes 10 minutes to complete and has a predicting accuracy of 71%.
“Our results encourage further studies utilizing multivariate approaches, as they emphasize that data mining and advanced statistics are auspicious in the quest for precision medicine in mental health,” the researchers wrote.
Kautzky A, Dold M, Bartova L, et al. Refining prediction in treatment-resistant depression: results of machine learning analyses in the TRD III sample. J Clin Psychiatry. 2018;79(1):16m11385.