Mood Disorders: Predictors of tDCS Response

Detecting reliable predictors of response to tDCS treatment for mood disorders could potentially facilitate treatment stratification and guide personalized treatment decisions, thereby enhancing treatment efficiency and reducing redundant treatments. Furthermore, hypothesis generation based on predictive information could streamline future research by narrowing the choice of multiple treatment parameters to the most promising targets. While attempts to identify such predictors have been made across multiple domains, these have mostly applied post-hoc exploratory statistics, that likely suffer from poor generalizability. As a result, objective response prediction is currently not feasible. New research strategies relying on artificial intelligence and statistical learning methodology represent promising means to achieve more reliable, clinically translatable predictive models. To be successful, these rely on multisite collaboration to enhance dataset sizes, should include data from naturalistic clinical settings, and embrace an open science framework. At the moment, heuristic clinical decision-making based on RCT results and clinical reasoning should guide tDCS treatment.