Development of a federated learning approach to predict acute kidney injury in adult hospitalized patients with COVID-19 in New York City

  1. Suraj K Jaladanki
  2. Akhil Vaid
  3. Ashwin S Sawant
  4. Jie Xu
  5. Kush Shah
  6. Sergio Dellepiane
  7. Ishan Paranjpe
  8. Lili Chan
  9. Patricia Kovatch
  10. Alexander W Charney
  11. Fei Wang
  12. Benjamin S Glicksberg
  13. Karandeep Singh
  14. Girish N Nadkarni

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Federated learning is a technique for training predictive models without sharing patient-level data, thus maintaining data security while allowing inter-institutional collaboration. We used federated learning to predict acute kidney injury within three and seven days of admission, using demographics, comorbidities, vital signs, and laboratory values, in 4029 adults hospitalized with COVID-19 at five sociodemographically diverse New York City hospitals, between March-October 2020. Prediction performance of federated models was generally higher than single-hospital models and was comparable to pooled-data models. In the first use-case in kidney disease, federated learning improved prediction of a common complication of COVID-19, while preserving data privacy.

Article activity feed

  1. ScreenIT

    SciScore for 10.1101/2021.07.25.21261105: (What is this?)

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

    Table 1: Rigor

    Ethicsnot detected.
    Sex as a biological variablenot detected.
    Randomizationnot detected.
    Blindingnot detected.
    Power Analysisnot detected.

    Table 2: Resources

    No key resources detected.

    Results from OddPub: Thank you for sharing your code.

    Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:
    Our study has several limitations. First, our data was limited to a single hospital system within New York City, which may limit generalizability to other geographical regions. Similarly, as the standard of care for hospitalized patients with COVID-19 has evolved after our data was collected, model performance should be validated with newer cohorts. Our models did not use inputs such as imaging and echocardiograms, and their exclusion may have hindered performance. We deliberately limited hyperparameter tuning of models to simulate a scenario where a federated model has to be urgently deployed at resource-constrained hospitals, such as during the COVID-19 pandemic. Allowing further optimization of the MLP structure, and techniques such as transfer learning may improve MLPfederated model performance.20 We examined only two classifiers; other approaches such as random forests or support vector machines may have yielded better results. Lastly, performance with the FL approach was generally statistically inferior to pooled dataset performance, and it remains to be explored if other approaches exist which can help bridge this gap while retaining the privacy advantages of FL. In summary, we demonstrate the utility of FL to improve prediction of key outcomes while maintaining privacy and confidentiality. We hope this will encourage the development of generalizable clinical models which would otherwise be hindered by inability to share patient-level data across institutional boundari...

    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.

    Results from scite Reference Check: We found no unreliable references.

    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.

  2. Version published to 10.1101/2021.07.25.21261105v1 on medRxiv