To evaluate whether wearable-derived sleep and activity features could complement baseline clinical variables for daily morning risk triage among psychiatric inpatients in a closed ward.
Approach:
Study Design: A prospective observational pilot study involving 87 psychiatric inpatients, with assessments using the Columbia-Suicide Severity Rating Scale (C-SSRS) and wearable devices for sleep/activity data.
Data Collection: Participants wore Fitbit Sense devices to collect passive sleep and activity data, which were analyzed alongside baseline clinical variables.
Model Development: An L1-penalized logistic regression (LASSO) model was developed to evaluate the performance of the multimodal fusion model against conventional assessment.
Key Findings:
The multimodal fusion model showed higher recall (0.560) and F2-score (0.548) compared to the conventional assessment model (0.289 recall, 0.300 F2-score).
The fusion model identified C-SSRS-positive records from patients with low admission scores using wearable-derived activity features.
Interpretation:
Limitations:
The study's findings are exploratory and hypothesis-generating, with substantial overlap in confidence intervals across models.
The sample size was limited, necessitating larger studies for validation.
Conclusion:
These exploratory findings suggest that wearable-derived sleep and activity features may provide complementary information for daily morning risk triage in psychiatric inpatients.
