transfer learning; co-data; prior information; ridge regression
Abstract :
[en] In many high-dimensional prediction or classification tasks, complementary data on the features are available, e.g. prior biological knowledge on (epi)genetic markers. Here we consider tasks with numerical prior information that provide an insight into the importance (weight) and the direction (sign) of the feature effects, e.g. regression coefficients from previous studies. We propose an approach for integrating multiple sources of such prior information into penalised regression. If suitable co-data are available, this improves the predictive performance, as shown by simulation and application. The proposed method is implemented in the R package `transreg' (https://github.com/lcsb-bds/transreg).
Research center :
- Luxembourg Centre for Systems Biomedicine (LCSB): Biomedical Data Science (Glaab Group)
Disciplines :
Mathematics Human health sciences: Multidisciplinary, general & others
Author, co-author :
Rauschenberger, Armin ; University of Luxembourg > Luxembourg Centre for Systems Biomedicine (LCSB) > Biomedical Data Science
Landoulsi, Zied ; University of Luxembourg > Luxembourg Centre for Systems Biomedicine (LCSB) > Bioinformatics Core
van de Wiel, Mark A.; Amsterdam University Medical Centers > Epidemiology and Data Science ; University of Cambridge > Medical Research Council Biostatistics Unit
Glaab, Enrico ; University of Luxembourg > Luxembourg Centre for Systems Biomedicine (LCSB) > Biomedical Data Science
External co-authors :
yes
Language :
English
Title :
Penalised regression with multiple sources of prior effects
Publication date :
2023
Journal title :
Bioinformatics
ISSN :
1367-4803
eISSN :
1367-4811
Publisher :
Oxford University Press, Oxford, United Kingdom
Volume :
in press
Peer reviewed :
Peer Reviewed verified by ORBi
Focus Area :
Systems Biomedicine
European Projects :
H2020 - 779282 - ERA PerMed - ERA-Net Cofund in Personalised Medicine
FnR Project :
FNR14599012 - Validating Digital Biomarkers For Better Personalized Treatment Of Parkinson'S Disease, 2020 (01/05/2021-30/04/2024) - Enrico Glaab
Name of the research project :
DIGIPD > Validating Digital Biomarkers For Better Personalized Treatment Of Parkinson’s Disease > 01/05/2021 > 30/04/2024 > 2020 U-AGR-7276 - NCER/23/16695277/CLINNOVA (01/01/2023 - 31/12/2025) - GLAAB Enrico
Funders :
FNR - Fonds National de la Recherche [LU] CE - Commission Européenne [BE] Union Européenne [BE]