Behavioral intervention therapies can improve symptoms of depression but do not work for all patients. To identify brain circuit changes that might underlie treatment response, Zhang et al. performed a longitudinal fMRI imaging study on patients with depression and comorbid obesity who received either problem-solving therapy (I-CARE) or usual care (U-CARE). They found that decreased cognitive control circuit activity measured at different time points over 24 months correlated with better treatment outcomes. A predictive model based on control circuit changes at 2 months outperformed a standard model based on demographic and clinical parameters. These results suggest an approach that could support treatment outcome predictions for depression, but further optimization and validation are needed. —Daniela Neuhofer