Mahnaz Olfati - Institute of Medical Science and Technology, Shahid Beheshti University, Tehran, Iran
Shahrooz Faghih Roohi - Institute of Medical Science and Technology, Shahid Beheshti University, Tehran, Iran
Fatemeh Samea - Institute of Medical Science and Technology, Shahid Beheshti University, Tehran, Iran
Masoud Tahmasian - Institute of Medical Science and Technology, Shahid Beheshti University, Tehran, Iran

IntroductionSleep disturbances are common symptoms in depression [۱]. Functional connectivity links mediate the effect of sleep quality on depressive problems [۲]. Moreover, the gray matter volume of the right insula mediates between depression/anxiety and sleep quality among college students [۳]. Nevertheless, the predictive role of brain structures in the association of sleep quality and depressive symptoms is still unclear. We aimed to identify whether sleep quality can predict depressive symptoms, as well as to assess the role of brain structures in the association between sleep quality and depressive symptoms in healthy subjects.MethodIn this study, we included ۱۱۰۱ participants from the human connectome project (HCP, ۵۹۸ females, ۲۲-۳۵ years). Depressive symptoms and sleep quality were assessed using the depressive problems portion of the Achenbach adult self-report and Pittsburgh sleep quality questionnaires, respectively. The ensemble machine learning algorithm was used for the prediction of depression based on sleep quality scores. Then, we used structural equation modeling to find the gray matter structure’s role in the relationship between sleep and depressive symptoms. We controlled for age and gender in all analyses.ResultsThe results revealed a significant correlation between sleep and depressive symptoms (r = ۰.۳۷, p < ۰.۰۰۱). The ensemble machine learning algorithm predicted depressive symptoms based on sleep quality (Accuracy = ۰.۸, Mean Square Error (MSE) = ۱۱.۴۱, Mean Absolute Error (MAE) = ۲.۵۳, ۹۵% Confidence Interval (CI) = ۳.۸۱_۴.۸۶, and 𝑅۲=۰.۴۶. (Figure ۱). After Bonferroni correction, we found two brain regions (the right somatomotory area and cerebellum) that had a significant correlation with depressive symptoms (p < ۰.۰۵). Mediation analysis showed that these regions are partial mediators on the effect of sleep quality on depressive symptoms (Table ۱).ConclusionsOur findings demonstrated that sleep quality scores predicted depressive symptoms via the right somatomotory area and cerebellum in healthy subjects.