Deep Learning-Based Prediction Model of Heart Failure with Improved Ejection Fraction

Authors:
Preprint
Publisher website
Google Scholar
Abstract
Background: This study developed a deep learning model to predict improvement of left ventricular ejection fraction in patients with heart failure. Methods: An internal database comprising clinical, laboratory and echocardiographic features obtained from the First Affiliated Hospital of Harbin Medical University was used to construct and train the deep learning model. Results: A total of 422 cases were included in this study. 122 (28.9%) were patients with HFimpEF, 300 (71.0%) were patients with HFrEF. Multivariable analyses showed that smaller baseline left atrial anterior-posterior diameter (LAD) and left ventricular end-diastolic diameter (LVEDD), higher baseline interventricular septal thickness at diastole (IVSD) and levels of prealbumin were the independent clinical predictors of LVEF improvement. Deep learning model demonstrated an overall predict accuracy of 96% in the validation set and 89% in the training set. Conclusions: Independent predictors of LVEF improvement were smaller baseline LVEDD, LAD, higher baseline IVSD and baseline levels of prealbumin. Our deep learning model had shown acceptable performance in predicting improvement of left ventricular ejection fraction in patients with heart failure. Registration: URL: https://www.clinicaltrials.gov ; Unique identifier: NCT06070506 .

This publication has 20 references indexed in Scilit: