Improving End-of-Life Screening in the Emergency Department With Collaborative Artificial Intelligence

The authors aimed to compare the effectiveness of the physician-answered surprise question (SQ) and the Geriatric End-of-Life Screening Tool (GEST), as well as a combined GEST+SQ model, in predicting 6-month mortality among older patients in the emergency department. Their findings indicated that while GEST outperformed SQ in terms of specificity and sensitivity, the combined model did not significantly enhance predictive discrimination but did improve calibration. The study suggests that a sequential screening approach could substantially reduce the screening burden on physicians while maintaining effective mortality risk assessment.