Predicting next-day suicidal urges, depression, and PTSD symptoms using ecological momentary assessments and passive sensing of sleep among military service members and veterans.

BACKGROUNDS: Suicide risk fluctuates rapidly, highlighting the importance of identifying risk factors for acute suicidal urges. This study examined whether nightly sleep measured actively via self-report ecological momentary assessment (EMA) and passively via wearable sensors predicted next-day suicidal urges, depression, and PTSD symptoms among military service members and veterans.

METHODS

Military service members and veterans with current suicidal ideation or a suicide attempt in the past month (N = 86) completed seven EMA surveys per day and wore a wearable device (Fitbit) for 28 days. Using multilevel models, we examined sleep assessed by EMA-only, wearable-only, and EMA + wearable as predictors of next-day suicidal urges, depression, and PTSD symptoms, and compared the predictive utility of the three approaches.

RESULTS

Participants completed 7898 EMA observations (48.9% adherence) and wore a wearable device (Fitbit) for 79.43% of the study period. More severe nightmares and poorer sleep quality assessed by EMA predicted next next-day suicidal urges (maximum and average), suicidal beliefs, depression, and PTSD symptoms. Wearable-assessed sleep duration deviation significantly predicted next-day maximum suicidal urges. Wearable-assessed sleep regularity index predicted next-day depression. EMA-only models consistently outperformed wearable-only models in predicting next-day suicide risks and mental health outcomes, and combining EMA and wearable demonstrated best model fit.

CONCLUSION

Our findings suggest that self-reported sleep via EMA has strong utility in predicting near-term suicidal risk, depression, and PTSD, while wearable devices can provide low-burden, supplemental information. Integrating wearables and EMA may enhance the prediction and inform just-in-time suicide interventions.