Detecting Anxiety States when Caring for People with Dementia

Area of Science:

• Physiological monitoring and computational modeling in healthcare.
• Wearable sensing technologies for mental health assessment.

Background:

• Caregiving, especially for individuals with dementia, is a significant source of stress, often leading to anxiety and depression.
• Wearable sensors offer a promising avenue for monitoring caregiver physiological data to detect and manage anxiety.

Purpose of the Study:

• To develop and evaluate a method for inferring caregiver anxiety states using physiological data.
• To specifically address the needs of caregivers for people with dementia.

Main Methods:

• A Markov chain model was developed to detect internal anxiety states.
• The model utilized Inter-beat Interval (IBI) data from 10 participants in a naturalistic experiment.
• Four states were classified: Relaxed, Arousing, Anxiety, and Relaxing.

Main Results:

• Visual analysis identified 823 state segments, with 'anxious' being the most frequent (410 segments).
• The Markov chain model achieved an average accuracy of 73.03% using 60-second IBI segments in a leave-one-out cross-validation.
• The model demonstrated effectiveness in classifying distinct anxiety-related states.

Conclusions:

• A novel Markov chain model effectively detects internal anxiety states in dementia caregivers.
• The model's performance is comparable to existing stress classification methods.
• This approach provides a foundation for developing technological interventions to support caregiver mental well-being.