Passive versus active night time phone use as markers of next day suicide risk: a high resolution screenome study

Seventy-nine adults with recent suicidal thoughts participated in a 28-day study combining ecological momentary assessment with Screenomics, a digital phenotyping method that captured over 7.5 million smartphone screenshots at 5-second intervals. We examined nighttime smartphone use through maximum phone-free intervals, sleep-window usage, and hourly patterns. Keyboard presence, detected via deep learning, served as a proxy for active engagement. Bayesian multilevel models revealed that late-night phone use (11PM-1AM) consistently predicted higher next-day passive/active suicidal ideation and planning. Participants with 7-9 hour phone-free gaps showed lowest ideation compared to 4-7 hour gaps, with no linear relationship. Keyboard activity predicted lower suicidal ideation when occurring during middle-night hours (1-5AM), self-reported sleep windows, and non-sleep periods. These timing-dependent, within-person effects suggest that when and how smartphones are used matters more than total usage. Findings enable personalized, real-time interventions that could support adaptive digital coping while flagging high-risk moments for suicide prevention.