Depression diagnosis depends on the transcripts obtained from clinical interviews during mental health assessment. This hybrid model proposes Decision Tree (DT) and Grey Wolf Optimizer (GWO) for feature selection to enhance depression detection. The proposed model makes use of the Decision Tree, thereby effectively grasping the temporal context embedded in clinical interview transcripts. It can detect significant linguistic features of depressive symptoms. This model will also include the integration of GWO to optimize feature selection for added strength and efficiency. In this way, a hybrid model featuring a high F-score macro of 0.83 could be derived, which proved its effectiveness in detecting depression with accuracy. This will contribute to the literature by enhancing computational methods for measuring mental health, and this model has immense potential to apply in clinical practice. Beyond academic research, the applications of such a hybrid model are huge and promising in a clinical setting. The model automates the diagnosis of depression to assist healthcare professionals in informed and timely intervention.