Introduction Sleep disorder is often the first symptom of age-related cognitive decline associated with Alzheimer’s disease (AD) observed in primary care. The relationship between sleep and early AD was examined using a patented sleep mattress designed to record respiration and high frequency movement arousals. A machine learning algorithm was developed to classify sleep features associated with early AD. Method Community-dwelling older adults (N = 95; 62–90 years) were recruited in a 3-h catchment area. Study participants were tested on the mattress device in the home bed for 2 days, wore a wrist actigraph for 7 days, and provided sleep diary and sleep disorder self-reports during the 1-week study period. Neurocognitive testing was completed in the home within 30-days of the sleep study. Participant performance on executive and memory tasks, health history and demographics were reviewed by a geriatric clinical team yielding Normal Cognition (n = 45) and amnestic MCI-Consensus (n = 33) groups. A diagnosed MCI group (n = 17) was recruited from a hospital memory clinic following diagnostic series of neuroimaging biomarker assessment and cognitive criteria for AD. Results In cohort analyses, sleep fragmentation and wake after sleep onset duration predicted poorer executive function, particularly memory performance. Group analyses showed increased sleep fragmentation and total sleep time in the diagnosed MCI group compared to the Normal Cognition group. Machine learning algorithm showed that the time latency between movement arousals and coupled respiratory upregulation could be used as a classifier of diagnosed MCI vs. Normal Cognition cases. ROC diagnostics identified MCI with 87% sensitivity; 89% specificity; and 88% positive predictive value. Discussion AD sleep phenotype was detected with a novel sleep biometric, time latency, associated with the tight gap between sleep movements and respiratory coupling, which is proposed as a corollary of sleep quality/loss that affects the autonomic regulation of respiration during sleep. Diagnosed MCI was associated with sleep fragmentation and arousal intrusion.

No abstract

No abstract

Introduction Late-life depression has been proposed as a precursor to amnestic Mild Cognitive Impairment (aMCI), the prodrome of Alzheimer’s disease. Both conditions are associated with sleep and cognitive problems. We hypothesized that MCI and current depressive symptoms would co-occur more frequently, but express distinct sleep phenotypes. Methods Independently living older adults (N=80), age 62-90 (M=71.78, SD=5.98), were recruited from a geriatric psychiatry clinic and the community for a home sleep study. A clinical decision board and neurocognitive battery were used to determine MCI status. Participants completed the CES-D and depression history interview where endorsement of current depression was considered positive. Sleep was examined with wrist actigraphy for 7 days. Pittsburgh Sleep Quality Index (PSQI), Epworth Sleepiness Scale (ESS), Stanford Sleepiness Scale (SSS) provided subjective sleep quality. Results Based on these criteria, 41.2% of the sample were determined to be MCI (n=33); the remainder were deemed normative for age (NC; n=47). Chi-square analyses showed a higher frequency of MCI were positive for current depression than expected (14.2%; p=0.017). Repeated-measures MANOVA, using current depression symptoms and MCI as factors, revealed MCI was associated with longer sleep latency (p=0.035) and wake bout time (p=0.039); whereas, current depression was associated with longer sleep latency, more fragmentation/WASO, and lower sleep efficiency (p’s<0.05), self-report of poorer daytime dysfunction (p=0.005), and greater daytime sleepiness (p=0.001). MCI x current depression interactions were found for sleep latency (p=0.029); and PSQI sleep disturbances (p=0.005) and sleep medication (p=0.025). Conclusion Despite distinct sleep disordered phenotypes, the interaction of MCI and current depression is associated with delayed sleep onset, use sleep medication and report of sleep disturbances. Support This project was sponsored by: NASA, Maine Space Consortium; AG 056176, AG 053164 Vice President for Research, U. Maine; Maine Technology Institute; DoD Phase I SBIR and R44AG059536-01 SBIR Phase II Award.