Predicting high anger intensity using ecological momentary assessment and wearable-derived physiological data in a...

Area of Science:

• Digital health interventions
• Wearable sensor technology
• Mental health research

Background:

• Digital technologies can predict mood and behavior, aiding digital health interventions.
• No prediction models exist for trauma-exposed populations using real-world data.

Purpose of the Study:

• To determine if wearable-derived physiological data can predict anger intensity in trauma-exposed adults.

Main Methods:

• Combined heart rate variability (commercial wearable stress score) with ecological momentary assessment (EMA) data over 10 days (n=84).
• Selected five summary measures from stress scores collected 10 minutes prior to each EMA using factor analysis.

Main Results:

• A logistic mixed-effects model showed high predictive performance (AUC 0.761–0.899).
• Findings suggest physiological data can predict anger intensity in this population.

Conclusions:

• Promising findings have implications for developing novel prediction and treatment approaches for posttraumatic mental health.
• Further replication with larger datasets is needed to validate predictive performance.