Sleep efficiency in community-dwelling persons living with dementia: exploratory analysis using machine learning.

Area of Science:

• Gerontology
• Neuroscience
• Biomedical Engineering

Background:

• Sleep disturbances are prevalent in community-dwelling older adults with dementia, negatively impacting health and increasing caregiver burden.
• Predicting and improving sleep efficiency is crucial for managing dementia symptoms and enhancing quality of life in home settings.

Purpose of the Study:

• To develop a predictive model for sleep efficiency in community-dwelling older adults with dementia.
• To identify key features associated with sleep efficiency in this population using machine learning.

Main Methods:

• An exploratory, observational study involving 69 older adults with dementia.
• Data collected via actigraphy, sweat patches for cytokines, and baseline surveys (diseases, medications, psychological/behavioral symptoms, functional status, demographics).
• Machine learning (CatBoost model) used to identify the best predictive model for sleep efficiency and its top 10 associated features.

Main Results:

• The CatBoost model was identified as the best predictor of sleep efficiency.
• Top predictive features included sleep regularity, number of medications, dementia medication, daytime activity, instrumental activities of daily living, neuropsychiatric inventory, hypnotics, occupation, tumor necrosis factor-alpha, and waking hour lux.

Conclusions:

• A robust predictive model for sleep efficiency was established for community-dwelling older adults with dementia.
• The findings underscore the need for individualized sleep interventions tailored to specific patient features, leveraging diverse data sources including IoT devices.