Machine learning for risk stratification in the emergency department (MARS-ED) study protocol for a randomized controlled pilot trial on the implementation of a prediction model based on machine learning technology predicting 31-day mortality in the emergency department

Comprehensive Summary

Van Dam et al. conducted a clinical trial to explore the clinical implications of a machine learning-based prediction model. The RISK index predicts 31-day mortality risk based on laboratory tests and clinical data. Prior studies have shown that the RISK index outperforms internal medicine physicians in mortality prediction. Participants were randomly assigned to control or intervention groups. Both groups received standard care, with the intervention group's physicians additionally receiving RISK index predictions. Attending physicians in the intervention group could incorporate RISK index results into their clinical decision-making at their discretion. This is a study protocol; results have not yet been published.

Outcomes and Implications

If effective, implementation of the RISK index in clinical settings may help practitioners better predict patients' 31-day mortality risk following emergency department admission. Earlier studies have shown that the RISK index outperforms physician prediction, suggesting its application could offer more standardized and objective risk assessment. More accurate risk estimation could enable hospitals to allocate resources and prioritize interventions more effectively.