Artificial Intelligence ECG Diastolic Dysfunction and Survival in Cardiac Intensive Care Unit Patients

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Left ventricular diastolic dysfunction (LVDD) is a significant predictor of mortality in cardiac intensive care units (CICU).
• An artificial intelligence-enhanced ECG (AIECG) algorithm has shown promise in predicting LVDD and mortality in general populations.
• The efficacy of AIECG in the specific context of CICU patients remains underexplored.

Purpose of the Study:

• To evaluate the association between AIECG-derived LVDD grade and mortality in patients admitted to a CICU.
• To determine if AIECG LVDD grading can predict in-hospital and 1-year mortality independently of traditional clinical variables and echocardiographic measurements.

Main Methods:

• A historical cohort study involving 11,868 adult patients admitted to a Mayo Clinic CICU between 2007 and 2018.
• AIECG was used to assign an LVDD grade (0-3) upon admission.
• In-hospital and 1-year mortality were assessed, with multivariable adjustments including transthoracic echocardiogram (TTE) data for elevated filling pressures (medial mitral E/e' ratio >15).

Main Results:

• Mortality rates increased with each higher AIECG LVDD grade.
• After multivariable adjustment, higher AIECG LVDD grades were significantly associated with increased in-hospital (aOR 1.22) and 1-year mortality (aHR 1.23).
• Patients with concordant AIECG (grade 2-3) and TTE (E/e' ratio >15) findings for elevated filling pressures exhibited the highest mortality risks (aOR 2.54, aHR 2.03).

Conclusions:

• AIECG LVDD grading is a strong, independent predictor of mortality in CICU patients.
• The AIECG algorithm provides valuable prognostic information beyond standard clinical and echocardiographic assessments.
• Identifying patients with concordant AIECG and TTE findings of elevated filling pressures can help stratify high-risk individuals in the CICU.