The following is a summary of “Development of early warning scores or alerting systems for the prediction of adverse events in psychiatric patients: a scoping review,” published in the October 2024 issue of Psychiatry by Velasquez et al. 

Adverse events (AEs) in psychiatric settings pose significant challenges for individuals and healthcare providers. Research on early warning scores (EWS) and tools to predict patient deterioration remains limited. 

Researchers conducted a retrospective review of predictive tools developed in psychiatric settings, comparing machine learning (ML) and traditional methodologies. 

They searched 3 databases—Ovid MEDLINE, PsycINFO, and Embase—from February to April 2023 using terms like “Early warning,” “Alerting tool,” and “Psychiatry.” After screening 1,193 studies, 9 met the inclusion and exclusion criteria for review, following the PICOS model, the Joanna Briggs Institute (JBI) Reviewer’s Manual, and PRISMA guidelines. 

The results showed that 9 studies developed predictive models for AEs in psychiatric settings, encompassing 41,566 participants. The studies utilized both ML and non-ML algorithmic approaches, with performance metrics, primarily AUC ROC, varying between 0.62 and 0.95. The best-performing validated model was the random forest (RF) ML model, which scored 0.87 with a sensitivity of 74% and a specificity of 88%. 

The study concluded that while few predictive models exist for AEs in psychiatric settings, both ML and non-ML algorithms demonstrate moderate to good performance, highlighting the need for further research on their feasibility and efficacy. 

Source: bmcpsychiatry.biomedcentral.com/articles/10.1186/s12888-024-06052-z