Left ventricular systolic dysfunction identification using artificial intelligence-augmented electrocardiogram in...

Area of Science:

• Cardiology
• Artificial Intelligence in Medicine
• Diagnostic Accuracy

Background:

• Artificial intelligence-augmented electrocardiograms (AI-ECG) demonstrate potential for identifying left ventricular systolic dysfunction (LVSD).
• Left ventricular systolic dysfunction (LVSD) is a critical condition requiring accurate and timely diagnosis.

Purpose of the Study:

• To evaluate the accuracy of AI-ECG in identifying LVSD in patients within a cardiac intensive care unit (CICU).
• To assess the performance of AI-ECG compared to transthoracic echocardiogram (TTE) for LVSD detection.

Main Methods:

• A cohort of 5680 unique Mayo Clinic CICU patients admitted between 2007 and 2018 were analyzed.
• Patients underwent both AI-ECG and transthoracic echocardiogram (TTE) within 7 days.
• The discrimination ability of AI-ECG for LVSD was determined using receiver-operator characteristic (ROC) curve analysis (Area Under the Curve - AUC).

Main Results:

• The AI-ECG achieved an AUC of 0.83 for discriminating LVSD in the CICU cohort.
• Overall accuracy for LVSD detection by AI-ECG was 76%, with 73% sensitivity and 78% specificity.
• Higher AI-ECG performance was observed in younger patients (<70 years), males, and those without acute coronary syndrome (ACS).

Conclusions:

• AI-ECG demonstrates very good discrimination for LVSD in critically ill CICU patients.
• The AI-ECG tool may be particularly effective for screening LVSD in younger males and patients without ACS.
• AI-ECG presents a potential alternative for LVSD identification in resource-limited settings where TTE is unavailable.