Sleep quality prediction in caregivers using physiological signals.

Area of Science:

• Biomedical Engineering
• Gerontology
• Psychology

Background:

• Caregivers of people with dementia (CPWD) face significant stress, leading to poor sleep quality.
• Monitoring CPWD sleep quality is crucial for assessing caregiver stress.
• Existing sleep monitoring methods like polysomnography are costly and disruptive.

Purpose of the Study:

• To develop a clinical decision support system for predicting CPWD sleep quality.
• To utilize physiological signals from wearable devices for sleep analysis.
• To offer a non-invasive and practical approach to monitoring caregiver sleep.

Main Methods:

• Employed an E4 wristband to collect physiological data: heart rate variability, electrodermal activity, body movement, and skin temperature.
• Developed a system to predict sleep quality based on trends in physiological signals during deep sleep.
• Utilized a random forest classifier for performance evaluation.

Main Results:

• The system achieved 75% accuracy in predicting sleep quality and 73% in predicting restfulness.
• Key features for accurate prediction included sleep efficiency and skin temperature.
• Demonstrated the feasibility of using wearable sensors for sleep analysis in CPWD.

Conclusions:

• Wearable sensor technology can effectively measure sleep quality and restfulness in CPWD.
• The proposed system offers a viable alternative to traditional sleep monitoring methods.
• This approach supports better stress assessment and management for caregivers.