Table 1 Summary of datasets and methods reproduced in this work.

From: Privacy-first health research with federated learning

Study Topic

Manuscript

Study Design

Unit of Analysis

N

Statistical Model

Additional Methods

Measure

Covariates

Heart Failure

Chicco & Jurman11

Cohort Study

Individual

299

Logistic Regression

N/A

AUC

12

Diabetes

Smith, et al.27

Cohort

Individual

768

Neural network with 1 hidden layer

N/A

AUC

8

MIMIC-III

Johnson et al.12

Database

Individual & Dirichlet grouping

53,423

Deep neural network

N/A

N/A

92

SARS-CoV-2

Rugge et al.13

Cohort Study

Individual & random size grouping

9275

Logistic Regression

N/A

Odds Ratio

3

Avian Influenza

Fiebig et al. 28

Case Series

Individual & country grouping

294

Logistic Regression

Forward/Backward Selection

Odds Ratio

4

Bacteraemia

Harris et al.29

Case Control

Individual

159

Logistic Regression

Forward/Backward Selection

Odds Ratio

12

Azithromycin

Oldenburg et al.30

Cluster Randomized Trial

Individual & community of residence grouping

1712

GLM with a Binomial response and the log link

Standard Errors Clustered

Risk Ratio

0

Tuberculosis

Ohene et al.31

Case Series

Individual & multi-center grouping

3342

Logistic Regression

N/A

Odds Ratio

3

  1. GLM Generalized Linear Model, AUC Area under the Receiver Operating Characteristic Curve.