Methodological Concerns in Spittal et al.’s Meta-Analysis of Machine Learning Algorithms for Suicide Risk

We raise methodological concerns regarding Spittal et al.'s systematic review and meta-analysis of machine learning algorithms for predicting suicidal behaviours (PLoS Medicine, 2025). Key issues include the absence of calibration assessment, reliance on diagnostic rather than prognostic frameworks, and use of QUADAS instead of PROBAST.