Abstract: FR-PO0020
Feasibility and Accuracy of Left Ventricular Ejection Fraction Assessment by Nephrology Professionals Using Artificial Intelligence
Session Information
- Artificial Intelligence and Digital Health at the Bedside
November 07, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Filardo-Zuniga, Marimar, Tecnologico de Monterrey - Campus Ciudad de Mexico, Mexico City, CDMX, Mexico
- Nieves Hernández, Luz Atenas, Tecnologico de Monterrey - Campus Ciudad de Mexico, Mexico City, CDMX, Mexico
- Hernandez-Garcia, Leslie Nahomi, Tecnologico de Monterrey - Campus Ciudad de Mexico, Mexico City, CDMX, Mexico
- Arvizu Hernández, Mauricio, Instituto Nacional de Ciencias Medicas y Nutricion Salvador Zubiran, Mexico City, CDMX, Mexico
Background
Evaluate the feasibility and accuracy of nephrologists measuring the left ventricular ejection fraction (LVEF) in patients with chronic kidney disease (CKD) and heart failure using portable ultrasound with artificial intelligence compared to formal echocardiography by cardiologists.
Methods
A retrospective descriptive study was conducted in the cardionephrology clinic of a tertiary hospital in Mexico City from April 1 to July 10, 2024. LVEF measurements by nephrologists using the EchoNous KOSMOS portable ultrasound with artificial intelligence were compared to formal echocardiography performed by cardiologist echocardiographers.
Results
Twenty-seven patients with both nephrologist-performed and prior formal LVEF measurements were included. Of these, 55.5% were male, with a mean age of 43.2±16.1 years. Overall, 92.5% were on renal replacement therapy: 62.9% on hemodialysis (mean 4 years), 29.6% on peritoneal dialysis (mean 3 years), one in pre-dialysis, and one post-transplant. The correlation between LVEF measurements by nephrologists and cardiologists was strong (r=0.825), with good agreement on Bland-Altman analysis. Mean LVEF was 49.5±11.6% (nephrologists) and 50.5±13.3% (echocardiography). LVEF classification was consistent: 28.5% reduced, 25.9% mildly reduced, and 51.8% preserved.
Conclusion
Nephrologists can accurately calculate LVEF using artificial intelligence in patients with CKD and heart failure, yielding comparable results to cardiologists. This demonstrates the feasibility of integrating nephrologists into the cardiac assessment of these patients.