Prediction of depressive symptoms severity based on sleep quality, anxiety, and brain: a machine learning approach across three cohorts

Background Depressive symptoms are rising in the general population, but their associated factors are unclear. Although the link between sleep disturbances and depressive symptoms severity (DSS) is reported, the predictive role of sleep on DSS and the impact of anxiety and the brain on their relationship remained obscure. Method Using three population-based datasets, we trained the machine learning models in the primary dataset (N = 1101) to assess the predictive role of sleep quality, anxiety, and brain structure and function measurements on DSS, then we tested our models’ performance in two independent datasets (N = 334, N = 378) to test the generalizability of our findings. Furthermore, we applied our machine learning model to a smaller longitudinal sample (N = 66). In addition, we performed a mediation analysis to identify the role of anxiety and brain measurements on the sleep quality-DSS link. Findings Sleep quality could predict individual DSS (r = 0.43, R[2][1] = 0.18, rMSE = 2.73), and adding anxiety, rather than brain measurements, strengthened its prediction performance (r = 0.67, R[2][1] = 0.45, rMSE = 2.25). Importantly, out-of-cohort validations in other cross-sectional datasets and a longitudinal sample provided robust results. Furthermore, anxiety scores (not brain measurements) mediated the association between sleep quality and DSS. Interpretation Poor sleep quality could predict DSS at the individual subject level across three cohorts. Anxiety symptoms not only increased the performance of the predictive model but also mediated the link between sleep and DSS. Evidence before this study Depressive symptoms are prevalent in modern societies, but their associated factors are less identified. Several studies suggested that sleep disturbance and anxiety are linked with depressive problems in the general population and patients with major depressive disorder. A few longitudinal studies and meta-analyses also suggested that sleep disturbance plays a key role in developing depressive problems and clinical depression. However, those original studies mainly used conventional group comparison statistical approaches, ignoring the inter-individual variability across participants. Moreover, their data were limited to a single cohort, limiting the generalizability of their findings in other samples. Thus, large-scale multi-cohort studies using machine learning predictive approaches are needed to identify the complex relationship between sleep quality, anxiety symptoms, and depressive symptoms at the individual subject level. We also focused on the neurobiological underpinning of their interplay. Added value of this study In this study, we used machine learning which enables individual-level predictions and can validate models on unseen data, thus providing a more robust analytical framework. This study used three independent cohorts, included a longitudinal sample, and performed careful complementary analyses to examine the robustness of our findings considering the impact of lifetime history of depression, effects of sleep-related questions of the depressive assessment, most important parameters of sleep quality in prediction of depressive symptoms severity, and testing the reverse direction i.e., predicting sleep quality based on depressive symptoms. We found that poor sleep quality could robustly predict depressive symptoms across three cohorts, but the reverse direction (prediction of sleep quality based on depressive symptoms) was less robust. Anxiety symptoms improved the performance of the predictive model and mediated the link between sleep and depressive symptoms. However, brain structure and function did not play an important role in their association. Our longitudinal data also highlighted the predictability of future depressive symptoms severity and the role of interventions (i.e., neurofeedback) in the prediction of future depressive symptoms based on sleep and anxiety. Implications of all the available evidence As depressive symptoms have a strong impact on public health, identifying their contributing factors such as poor sleep and anxiety is critical to decrease the burden of depressive symptoms and/or design better therapeutical approaches at the individual subject level. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study did not receive any funding ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: We used the public datasets from the Human Connectome Project () and the enhanced Nathan Kline Institute-Rockland sample (eNKI) (). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present study are available upon reasonable request to the authors [1]: #ref-2