Application of Multi Task Analysis Model Based on EEG Features in the Study of the Correlation between Depression and Sleep Disorders in College Students

With the improvement of multi-modal technology, research on depression recognition combining multiple biological signal features has been increasing. Electroencephalogram signals reflect brain activity, and by analyzing their characteristics, the accuracy and stability of depression recognition can be effectively improved. Therefore, this study proposes a correlation analysis between depression and sleep disorders in college students based on electroencephalogram feature analysis, innovatively combining the analysis of the relationship between brain functional connectivity and brain structural connectivity, and using this for sleep staging. The results demonstrated that the accuracy of all research methods was relatively high, generally above 80%. The F1 score of additive fusion was generally higher than that of multiplicative fusion in all subjects, except for subject n10, the F1 score of additive fusion was at least 0.1% higher than that of multiplicative fusion. The lowest F1 score of patient subjects appeared in the delta frequency band, at 80.27%, while the lowest F1 score of healthy subjects appeared in the sigma frequency band, at 79.86%, with little difference between the two. Patients with depression and sleep disorders among college students had a low score of 26 on the Quality of Life Scale for Children and Adolescents. There was a relation between sleep disorders and depression among college students. This study improved the accuracy and stability of depression recognition by integrating electroencephalogram features. This offers a new approach and method for the early detection and diagnosis of depression, and provides important basis for further research on the pathogenesis and treatment of depression.