dsMTL client site functions

dsMTL server site functions

We propose dsDid, a federated learning package implemented in DataSHIELD with a federated version of the DID approach of Callaway and Sant'Anna (2022) at its core. It allows for the federated estimation of treatment effects per period and the corresponding federated uncertainty quantification.

An Algorithm for Privacy-preserving Efficient Aggregation of Longitudinal Data

Part of a research project for Maastricht University (semester 1). Contributors: Filippo Molino, Matteo Marrocco, Juliette Maes, Ron Dsilva, Ahmeduallah Khan, Leo Paggen.

Federated learning on radiomics data using grid logistic ordinal regression.

End to end sim for secure GWAS and PGS benchmarking. Uses 1000G based genotypes, with synthetic T2D and BMI traits from the PGS catalog. Compares baseline, differential privacy, and federated learning results. Data and PGS files not in repo, fetch from IGSR and the PGS Catalog. R code with bash/Slurm, and built to run on an HPC cluster. WIP.

This is an R package for feature alignment issues in vertical federated learning