Sensing a problem: Proof of concept for characterizing and predicting agitation

Area of Science:

• Gerontology
• Biomedical Engineering
• Data Science

Background:

• Agitation in dementia patients presents significant management challenges and caregiver stress.
• Current agitation assessment methods (self-report, brief observation) lack sensitivity and comprehensiveness.
• Early detection and identification of agitation triggers are crucial for effective care.

Purpose of the Study:

• To demonstrate the feasibility of a continuous monitoring system for characterizing and predicting agitation in dementia patients.
• To establish a proof of concept for using sensor data to understand agitation patterns.
• To identify potential behavioral and environmental precipitants of agitation.

Main Methods:

• Utilized a system of unobtrusive behavioral sensors (motion, door, actigraphy, bed pressure) and environmental sensors (temperature, light, sound, humidity).
• Collected continuous data over 138 days for a participant in a memory care facility.
• Compared features from agitated and non-agitated nursing shifts using t-tests.

Main Results:

• Participant activity metrics (e.g., transitions, sleep scores, activity counts) significantly correlated with nighttime agitation (P < 0.05).
• Environmental variables, specifically humidity, also showed a significant correlation with nighttime agitation (P < 0.05).
• Increased participant activity was observed in the evenings preceding agitated nights.

Conclusions:

• A sensor-based platform for continuous, unobtrusive monitoring of dementia patients and their environment is feasible.
• The system shows promise for characterizing agitation episodes.
• The technology can aid in identifying behavioral and environmental factors that precipitate agitation.