Generalized promotion time cure model: A new modeling framework to identify cell-type-specific genes and improve survival prognosis

2025-08-31

AI Summary

Existing statistical models struggle to jointly integrate individual-level survival data, tissue composition at the multicellular level, and single-cell omics features for identifying cell-type-specific prognostic genes. To address this, we propose the Bayesian Generalized Promotion Time Cure Model (BGPTM), the first framework enabling integrative modeling across three biological scales—individual survival outcomes, cell-type abundances, and single-cell gene expression—while supporting high-dimensional variable selection and survival prediction. BGPTM incorporates a Bayesian sparse prior, a cell-type-specific effect structure, and a time-dependent cure mechanism. In both simulation studies and real-world cancer datasets, BGPTM significantly improves covariate identification accuracy and prognostic prediction performance, achieving an average 0.08–0.12 increase in C-index. The model provides an interpretable, scalable statistical framework for dissecting tumor microenvironment-driven heterogeneity in patient prognosis.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Bayesian Learning
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Single-cell technologies provide an unprecedented opportunity for dissecting the interplay between the cancer cells and the associated tumor microenvironment, and the produced high-dimensional omics data should also augment existing survival modeling approaches for identifying tumor cell type-specific genes predictive of cancer patient survival. However, there is no statistical model to integrate multiscale data including individual-level survival data, multicellular-level cell composition data and cellular-level single-cell omics covariates. We propose a class of Bayesian generalized promotion time cure models (GPTCMs) for the multiscale data integration to identify cell-type-specific genes and improve cancer prognosis. We demonstrate with simulations in both low- and high-dimensional settings that the proposed Bayesian GPTCMs are able to identify cell-type-associated covariates and improve survival prediction.

Problem

• Integrates multiscale survival, composition, and omics data
• Identifies cell-type-specific genes predictive of cancer survival
• Improves cancer prognosis through Bayesian generalized promotion time models

Innovation

• Bayesian generalized promotion time cure models
• Multiscale data integration framework
• Cell-type-specific gene identification