dsMTL (Federated Multi-Task Learning based on DataSHIELD) provided federated, privacy-preserving MTL analysis. dsMTL was developed based on DataSHIELD, an ecosystem supporting the federated analysis of sensitive individual-level data that remains stored behind the data owner’s firewall throughout analysis. Multi-task Learning (MTL) aimed at simultaneously learning the outcome (e.g. diagnosis) associated patterns across datasets with dataset-specific, as well as shared, effects. MTL has numerous exciting application areas, such as comorbidity modeling, and has already been applied successfully for e.g. disease progression analysis.
dsMTL contained these algorithms:
This version of dsMTL can be found here:
Client: GitHub - transbioZI/dsMTLClient: dsMTL client site functions
Server: GitHub - transbioZI/dsMTLBase: dsMTL server site functions
In this version, we added two significant features.
1, We added a new algorithm dsLassoCov which allowed the control of covariates during the model training. The manuscript will be attached as soon as possible. The algorithm codes is here: dsMTLClient/ds.LassoCov.R at main · transbioZI/dsMTLClient · GitHub
2, we re-create all algorithms’ tutorial and test files with the R package DSLite. This is a simulation software of DataSHIELD servers in one machine. The testing files can be found here: dsMTLClient/tests/DSLite at main · transbioZI/dsMTLClient · GitHub
Dr. Han Cao