Comprehensive Summary
This study used data from a Slovenian nationwide community sample (N = 2,989) via online questionnaire to build and validate machine-learning models for indirect screening of both suicidal ideation and moderate-to-severe depression, including in individuals with subthreshold insomnia. The predictive performance of the SI model was 0.78 in the insomnia group and 0.80 in the non-insomnia group; for the depression model it was 0.79 and 0.82 respectively.
Outcomes and Implications
The findings suggest that ML-based indirect screening tools (using socio-demographics, coping strategies, behavioral changes) can effectively detect SI and depression even in people with sleep complaints, offering a feasible, time-efficient mechanism to identify at-risk individuals and enabling earlier intervention in settings where sleep disturbances are a common presenting complaint.