Clinical AdvisorAI May Help Predict In-Hospital Cardiac Arrest

The integration of AI, specifically a multilayer perceptron (MLP) model, into emergency department protocols could significantly enhance the prediction accuracy of in-hospital cardiac arrest (IHCA), surpassing traditional methods like the Modified Early Warning Score (MEWS).

Artificial intelligence (AI) has shown promising results in predicting in-hospital cardiac arrest (IHCA) in the emergency department (ED) more accurately than conventional screening protocols. A study presented at the American Academy of Physician Associates (AAPA 2024) annual meeting highlighted the superior performance of a multilayer perceptron (MLP) model compared to the Modified Early Warning Score (MEWS), suggesting significant potential for AI integration to improve clinical decision-making, triage, and resource optimization in emergency settings.

Key Points:

• MLP outperformed the Modified Early Warning Score (MEWS), a traditional method used since 1997, in predicting IHCA.
• MEWS has limitations, including low sensitivity and a high rate of false positives.
• The study included 448,972 patients from Beth Israel Deaconess Medical Center, Boston, spanning 2011 to 2019, excluding those under 18 and those presenting in active cardiac arrest.
• The MLP model utilized clinical features such as age, gender, and vital sign changes before IHCA.
• MLP achieved an AUROC score of 0.77, compared to 0.683 for MEWS, indicating better predictive accuracy.
• Integration of AI models like MLP can significantly enhance decision-making capabilities for PAs and other healthcare professionals in the ED.
• Annually, more than 290,000 IHCA cases occur in the US, with 10%-20% occurring in the ED, highlighting the critical need for improved prediction methods.
• The use of AI can optimize resource allocation and improve patient outcomes by enabling timely clinical decisions.

In-hospital cardiac arrest occurs in more than 290,000 adults each year in the United States. Cohort data from the United States indicate that the mean age of patients with in-hospital cardiac arrest is 66 years, 58% are men, and the presenting rhythm is most often (81%) nonshockable (ie, asystole or pulseless electrical activity). (JAMA)

More in Cardiology