PROJECT SUMMARY Nearly every psychiatric, neurodegenerative, and neurodevelopmental disorder has sleep dysfunction as a symptom. However, the specificity of biomarkers of sleep dysfunction for a given disease process is poorly un- derstood, due to the extreme heterogeneity observed within healthy populations and clinical cohorts. Of particular importance to neurological biomarker identification are sleep spindles—waxing-waning EEG waveforms of ~15Hz oscillatory activity that define NREM Stage 2 (N2) sleep. Spindle activity has been linked to memory consolidation during sleep and associated with numerous psychiatric, neurodegenerative, and neurodevelop- mental diseases, including Alzheimer's disease, trisomy 21, schizophrenia, and autism, as well as with natural aging. Recent work has shown that spindles have trait-like intrasubject stability, but also vast intersubject heter- ogeneity—the extent of which has yet to be characterized. Without characterizing spindle variability, we cannot understand the specificity of sleep EEG biomarkers. In this study, we develop a robust mathematical framework for understanding spindle temporal dynamics. Using this framework, we will characterize the variability in spindle activity across a multi-ethnic, heterogenous population of thousands of subjects across a wide age range. We will then use these normative distributions to evaluate the specificity of established and novel biomarkers of spindle dysfunction in trisomy 21 and Alzheimer's disease and mild cognitive impairment cohorts. In doing so, we fill an important gap essential to Alzheimer's disease biomarker development, providing a context and spec- ificity lacking in current approaches.

PROJECT NARRATIVE Nearly every psychiatric, neurodegenerative, and neurodevelopmental disorder has sleep dysfunction as a symptom, but how specific is a given symptom of pathological sleep to a given disorder? We will develop a powerful new approach for modeling the dynamics of sleep spindles—brain waveforms occurring during sleep, changes in which are associated with disease states and natural aging—which we will use to characterize vari- ability in ~18,000 subjects, as well as in trisomy 21 and Alzheimer's disease and mild cognitive impairment cohorts. By incorporating broad population variability into the analysis of pathological sleep, we fill an important gap essential to sleep phenotyping, providing a context and specificity lacking in current approaches to Alz- heimer's disease biomarker detection.