Search papers, labs, and topics across Lattice.
The paper introduces CURE, a novel framework for counterfactual time-to-event survival prediction that leverages multimodal data integration and latent subgroup retrieval to optimize individualized survival outcomes. CURE employs cross-attention to fuse clinical, paraclinical, demographic, and multi-omics data, and uses a mixture-of-experts architecture to refine complex multi-omics signals. Experiments on METABRIC and TCGA-LUAD datasets demonstrate that CURE outperforms existing survival analysis methods, as measured by the Time-dependent Concordance Index and Integrated Brier Score.
CURE achieves state-of-the-art survival prediction by implicitly retrieving patient-specific latent subgroups, capturing both baseline survival dynamics and treatment-dependent variations from multimodal data.
This paper tackles the problem of time-to-event counterfactual survival prediction, aiming to optimize individualized survival outcomes in the presence of heterogeneity and censored data. We propose CURE, a framework that advances counterfactual survival modeling via comprehensive multimodal embedding and latent subgroup retrieval. CURE integrates clinical, paraclinical, demographic, and multi-omics information, which are aligned and fused through cross-attention mechanisms. Complex multi-omics signals can be adaptively refined using a mixture-of-experts architecture, emphasizing the most informative omics components. Building upon this representation, CURE implicitly retrieves patient-specific latent subgroups that capture both baseline survival dynamics and treatment-dependent variations. Experimental results on METABRIC and TCGA-LUAD datasets demonstrate that proposed CURE model consistently outperforms strong baselines in survival analysis, evaluated using the Time-dependent Concordance Index ($C^{td}$) and Integrated Brier Score (IBS). These findings highlight the potential of CURE to enhance multimodal understanding and serve as a foundation for future treatment recommendation models. All code and related resources are publicly available to facilitate the reproducibility https://github.com/L2R-UET/CURE.
