Dynamic facial features in positive-emotional speech for identification of depressive tendencies

Abstract

Depressive symptoms in young people may persist into adulthood and develop into depression. Early screening of depressive tendencies in university students helps to reduce the number and intensity of their depressive episodes. Automatic prediction of depression from visual clues has gained increasing interest in recent years. This study proposes a deep learning model for depressive tendencies prediction, which extracts and fuses the dynamic facial features when participants were uttering positive text materials. The effectiveness of the proposed method is validated on our original multimodal behavioral dataset of Chinese university students. The results demonstrated that dynamic facial expression features can potentially reveal depressive tendencies. The average detection accuracy of individually uttered sentence is 67.32%, increasing to 71.4% when multiple sentences were fused.