Leveraging Sleep-Circadian Signatures to Predict Substance Use Outcomes in Adolescence

University Of Pittsburgh At Pittsburgh, Pittsburgh PA

Investigators

Linked publications, trials & patents

Abstract

PROJECT SUMMARY PROJECT 2 Adolescence heralds an alarming escalation in substance use (SU), especially nicotine and cannabis. SU is often initiated in middle adolescence and accelerates in late adolescence, making early adolescence (11-13yr) is a crucial window for detection and early intervention. Machine learning models to identify at-risk youth are expanding but have had only modest practicality and predictive value. Moreover, even if elevated SU risk is present (e.g., via maladaptive reward and cognitive control processes), early adolescents may not yet exhibit a level of SU that would trigger intervention. Modifiable SU risk factors should be incorporated into risk stratification models to enhance screening, improve prediction, and guide preventive care. Circadian rhythm and sleep disturbances present a collection of such risk factors, as they have strong ties to SU risk. Moreover, their effects on intermediary markers of SU risk in adolescence – disrupted reward and cognitive control systems—provides a plausible mechanistic substrate. Despite advances in our understanding and measurement of sleep-circadian health, our ability to use these data to identify at-risk youth is limited because: (1) the heterogeneity of sleep- circadian characteristics hinders the differentiation of normative versus at-risk patterns; (2) the multidimensionality and multimodality of sleep-circadian data has led to inconsistencies regarding which features and modalities are critical for screening; and (3) large samples with multimodal sleep-circadian data are rare, yet necessary to develop and validate screening algorithms. The primary goal of the Center for Adolescent Rhythms, Reward and Sleep (CARRS-2) is to understand how sleep and circadian rhythms during adolescence lead to increased vulnerability for SU. In CARRS-2 Project 2 (P2), we will adapt and apply a computational sleep health framework to identify multimodal sleep-circadian signatures, link them to prospective changes in SU risk, and develop practical screening algorthms. To do so, we will leverage the NIDA-funded Adolescent Brain Cognitive Development (ABCD) study: the largest longitudinal study of child development in the United States, in which about 4,800 children followed between ages 11-19yr are completing multimodal measures of SU, reward, cognitive control, and sleep-circadian characteristics (self-report, parent-report, FitBit). In ABCD, we will develop and validate multidimensional and multimodal sleep signatures (Aim 1) and practical and scalable SU risk stratification algorithms (Aim 2). Aims 1-2 will employ a sequential, multimodal machine learning workflow to establish the extent to which each data modality can incrementally inform sleep-circadian signatures and identify youth at greatest SU risk. In Aim 3, we will externally validate signatures and algorithms from Aims 1-2 using a one-of-a-kind harmonized University of Pittsburgh adolescent cohort. These aims will be accomplished via heavy integration with the Center Cores. Our long-term goal is to use modifiable sleep- circadian signatures to inform SU screening, risk stratification, and developmentally appropriate interventions.