Deep-learning-assisted analysis of echocardiographic videos improves predictions of all-cause mortality.

Journal: Nature biomedical engineering
Published Date: Feb 8, 2021

Abstract

Machine learning promises to assist physicians with predictions of mortality and of other future clinical events by learning complex patterns from historical data, such as longitudinal electronic health records. Here we show that a convolutional neural network trained on raw pixel data in 812,278 echocardiographic videos from 34,362 individuals provides superior predictions of one-year all-cause mortality. The model's predictions outperformed the widely used pooled cohort equations, the Seattle Heart Failure score (measured in an independent dataset of 2,404 patients with heart failure who underwent 3,384 echocardiograms), and a machine learning model involving 58 human-derived variables from echocardiograms and 100 clinical variables derived from electronic health records. We also show that cardiologists assisted by the model substantially improved the sensitivity of their predictions of one-year all-cause mortality by 13% while maintaining prediction specificity. Large unstructured datasets may enable deep learning to improve a wide range of clinical prediction models.

Authors

  • Alvaro E Ulloa Cerna
    Department of Translational Data Science and Informatics, Geisinger, Danville, PA, USA.
  • Linyuan Jing
    Department of Imaging Science and Innovation, Geisinger, Danville, Pennsylvania.
  • Christopher W Good
    Department of Cardiology, Geisinger, Danville, Pennsylvania.
  • David P vanMaanen
    Department of Translational Data Science and Informatics, Geisinger, Danville, PA, USA.
  • Sushravya Raghunath
    Department of Translational Data Science and Informatics, Geisinger, Danville, PA, USA.
  • Jonathan D Suever
    Department of Translational Data Science and Informatics, Geisinger, Danville, PA, USA.
  • Christopher D Nevius
    Department of Translational Data Science and Informatics, Geisinger, Danville, PA, USA.
  • Gregory J Wehner
    Department of Biomedical Engineering, University of Kentucky, Lexington, Kentucky.
  • Dustin N Hartzel
    Phenomic Analytics and Clinical Data Core, Geisinger, Danville, PA, USA.
  • Joseph B Leader
    Phenomic Analytics and Clinical Data Core, Geisinger, Danville, PA, USA.
  • Amro Alsaid
    Heart Institute, Geisinger, Danville, PA, USA.
  • Aalpen A Patel
    Department of Radiology, Geisinger, Danville, PA, USA.
  • H Lester Kirchner
    Department of Population Health Sciences, Geisinger, Danville, PA, USA.
  • John M Pfeifer
    Heart and Vascular Center, Evangelical Hospital, Lewisburg, PA, USA.
  • Brendan J Carry
    Heart Institute, Geisinger, Danville, Pennsylvania.
  • Marios S Pattichis
    Department of Electrical and Computer Engineering, University of New Mexico, Albuquerque, NM, United States.
  • Christopher M Haggerty
    IT Data Science, NewYork-Presbyterian Hospital, New York, New York, USA.
  • Brandon K Fornwalt
    Department of Imaging Science and Innovation, Geisinger, Danville, Pennsylvania; Department of Biomedical Engineering, University of Kentucky, Lexington, Kentucky; Department of Radiology, Geisinger, Danville, Pennsylvania. Electronic address: bkf@gatech.edu.

Keywords

External Resources

Popular Topics
Recent Journals