Dynamic prediction of inpatient self-harm and suicide attempts using machine learning: Toward data-driven clinical monitoring.

The improved performance of prediction models leveraging dynamic (daily) self-reported data could support just-in-time alerts for clinical staff. Further research is needed to identify salient markers for when the risk of self-harm and suicide attempts may be heightened to further enhance prediction accuracy. (PsycInfo Database Record (c) 2025 APA, all rights reserved).