This study, conducted by Huang et al., developed machine-learning models to predict short-term, mid-term, and long-term responses to antidepressants when using electroencephalographic (EEG) data. Seventy seven patients diagnosed with major depressive disorder (MDD) had their EEG data recorded before medication initiation and one week after medication initiation, along with seventeen demographic and behavioral variables also recorded. HAMD scores were the primary form of measurement of depression severity and this label became a learning target for training the machine learning models. All of these features and labels were input into nineteen machine learning classifiers for a performance assessment using leave-one-out cross-validation (LOOCV). A statistical analysis was also performed to understand the factors that influence treatment response and model performance. Results showed that the highest accuracy for the first stage of treatment at week 4, the second stage of treatment at week 6, and the third stage of treatment at week 8 was 0.831, 0.733, and 0.800, respectively. The Phase Lag Index (PLI) and Weighted Phase Lag Index (wPLI) had the most significant effect on accuracy, while including clinical features decreased the performance. This project confirmed that functional connectivity features from EEG are valuable for predicting responses to antidepressant treatment.