Transfer learning with chest X-rays for ER patient classification

  1. Jonathan Stubblefield
  2. Mitchell Hervert
  3. Jason L. Causey
  4. Jake A. Qualls
  5. Wei Dong
  6. Lingrui Cai
  7. Jennifer Fowler
  8. Emily Bellis
  9. Karl Walker
  10. Jason H. Moore
  11. Sara Nehring
  12. Xiuzhen Huang

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

One of the challenges with urgent evaluation of patients with acute respiratory distress syndrome (ARDS) in the emergency room (ER) is distinguishing between cardiac vs infectious etiologies for their pulmonary findings. We conducted a retrospective study with the collected data of 171 ER patients. ER patient classification for cardiac and infection causes was evaluated with clinical data and chest X-ray image data. We show that a deep-learning model trained with an external image data set can be used to extract image features and improve the classification accuracy of a data set that does not contain enough image data to train a deep-learning model. An analysis of clinical feature importance was performed to identify the most important clinical features for ER patient classification. The current model is publicly available with an interface at the web link: http://nbttranslationalresearch.org/ .

Article activity feed

  1. Version published to 10.1038/s41598-020-78060-4
  2. ScreenIT

    SciScore for 10.1101/2020.04.11.20062091: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    NIH rigor criteria are not applicable to paper type.

    Table 2: Resources

    Software and Algorithms
    SentencesResources
    We used the Python implementation [11].
    Python
    suggested: (IPython, RRID:SCR_001658)

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

    Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:
    Future Performance Expectations: The main limitation on this model’s current performance is the relatively small number of example cases. The dataset of 171 patients is far below an ideal number for training. However, we are continuing to expand our dataset. As our dataset grows, we expect significant performance improvements. We are also exploring new image model formulations that make use of “localization” annotations we were able to collect on our dataset. These annotations should allow us to provide addition feedback to the image model to serve as a forcing function for an attention mechanism. With an updated model and by expanding our dataset to hundreds of cases, we expect accuracy to make significant improvements in performance. Future Task Expansion: This project was started before the recent SARS-CoV-2 pandemic. As we move forward with development, we will explore upgrading the model to include a SARS-CoV-2 specific classification with COVID-19 patients data. This would allow physicians to use the same software to diagnose cases of SARS-CoV-2 pneumonia. We expect our model to be able to perform this task with a high accuracy as other research teams have had success with this problem [16]. This would also support our goal of improving antibiotic stewardship among physicians as SARS-CoV-2 pneumonia does not benefit from antibiotic therapy [11]. Future Research: In our next phase of research, we will allow our collaborating resident physicians to apply the model to new ...

    Results from TrialIdentifier: No clinical trial numbers were referenced.

    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

    Results from JetFighter: We did not find any issues relating to colormaps.

    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  3. Version published to 10.1101/2020.04.11.20062091v1 on medRxiv