Mortality Risk Stratification Utilizing Artificial Intelligence Electrocardiogram for Hyperkalemia in Cardiac Intensive Care Unit Patients

Background: Hyperkalemia has been associated with increased mortality in cardiac intensive care unit (CICU) patients. An artificial intelligence (AI) enhanced electrocardiogram (ECG) can predict hyperkalemia, and other AI-ECG algorithms have demonstrated mortality risk-stratification in CICU patients.

Objectives: The authors hypothesized that the AI-ECG hyperkalemia algorithm could stratify mortality risk beyond laboratory serum potassium measurement alone.

Methods: We included 11,234 unique Mayo Clinic CICU patients admitted from 2007 to 2018 with a 12-lead ECG and blood potassium (K) level obtained at admission with K ≥5 mEq/L defining hyperkalemia. ECGs underwent AI evaluation for the probability of hyperkalemia (probability >0.5 defined as positive). Hospital mortality was analyzed using logistic regression, and survival to 1 year was estimated using Kaplan-Meier and Cox analysis.

Results: In the final cohort (n = 11,234), the mean age was 69.6 ± 10.5 years, 37.8% were females, and 92.4% were White. Chronic kidney disease was present in 20.2%. The mean laboratory potassium value for the cohort was 4.2 ± 0.3 mEq/L. The AI-ECG predicted hyperkalemia in 33.9% (n = 3,810) of CICU patients and 12.9% (n = 1,451) of patients had laboratory-confirmed hyperkalemia (K ≥5 mEq/L). In-hospital mortality increased in false-positive, false-negative, and true-positive patients, respectively (P < 0.001), and each of these patient groups had successively lower survival out to 1 year.

Conclusions: AI-ECG-based prediction of hyperkalemia, even with a normal laboratory potassium value, was associated with higher in-hospital mortality and lower 1-year survival in CICU patients. This study demonstrated that AI-ECG probability of hyperkalemia may enable rapid individualized risk stratification in critically ill patients beyond laboratory value alone.