Machine learning in the prediction of depression treatment outcomes: a systematic review and meta-analysis

The negative relationship between study quality and prediction accuracy, combined with a lack of independent replication, invites caution when evaluating the potential of machine learning applications for personalizing the treatment of depression.