Abstract
In the era of big data, the simultaneous analysis of multiple high-dimensional, heavy-tailed datasets has become essential. Integrative analysis offers a powerful approach to combine and synthesize information from these various datasets, and often outperforming traditional meta-analysis and single-dataset analysis. In this paper, we introduce a novel high-dimensional integrative quantile regression that can accommodate the complexities inherent in multi-dataset analysis. A contrast penalty that smooths regression coefficients is introduced to account for across-dataset structures and improve variable selection. To ease the computational burden associated with high-dimensional quantile regression, a new algorithm is developed that is effective at computing solution paths and selecting significant variables. Monte Carlo simulations demonstrate its competitive performance. Additionally, the proposed method is applied to data from the China Health and Retirement Longitudinal Study, illustrating its practical utility in identifying influential factors affecting support income for the elderly. Findings indicate that adult children's individual characteristics and emotional comfort are primary factors of support income, and the extent of their impact varies across regions.