joinet

joinet implements multivariate elastic net regression by integrating multivariate lasso and ridge within a stacked generalization framework to predict correlated outcomes in high-dimensional clinical and genomic datasets.

Key Features:

• Multivariate Regression: Performs regression for multiple correlated target variables simultaneously to leverage outcome correlations.
• Stacked Generalization: Combines predictions from multiple models via stacked generalization to improve accuracy and robustness.
• Elastic Net Regression: Applies an elastic net penalty combining lasso (L1) and ridge (L2) to manage multicollinearity and aid interpretability.
• High-Dimensional Data Handling: Tailored for high-dimensional datasets and provides a single estimate for each input–output effect.
• Predictive Performance: Evaluated with rigorous simulation studies demonstrating competitive performance against other multivariate regression methods.

Scientific Applications:

• Biomedical and Clinical Research: Predicts multiple correlated outcomes in biomedical and clinical studies.
• Clinical and Genomic Data Analysis: Applied to clinical and genomic data to predict motor and non-motor symptoms in Parkinson's disease patients.
• Personalized Medicine and Complex Trait Analysis: Supports modeling of correlated phenotypes for personalized medicine and complex trait analyses.

Methodology:

Integrates multivariate lasso and ridge within an elastic net framework using stacked generalization to produce unified input–output effect estimates and is evaluated via simulation studies.