(1 Jilin Chemical Industry Hospital, Jilin City 132022, China;2. School of Information and Control, Jilin University of Chemic
Technology , Jilin City 132022,China)
Abstract: Breast cancer, as one of the most prevalent malignancies among women worldwide, exhibits significant heterogeneity in patient survival prognosis. This study proposes the DAIF-Cox model(Denoising Autoencoder with lterative Fusion and Cox Analysis) for breast cancer survival risk assessment, utilizing genomic data and whole-slide pathological image data from the TCGA database. The model employs a denoising autoencoder for feature compression, integrates an iterative attention mechanism for multimodal fusion, and ultimately constructs a Cox proportional hazards model for survival risk assessment.Experimental results demonstrate that the multimodal-fused DAlF-Cox model significantly outperforms unimodal models in survival prediction. Furthermore, ten key genes closely associated with breast cancer prognosis were successfully identified using XGBoost and Lasso regression methods, providing reliable candidates for prognostic gene analysis in breast cancer.
张 军, 吴昊诚, 韩 波, 尤涛, 魏子瑁, 冯欣, 辛瑞昊. 基于多模态融合的乳腺癌预后风险预测模型#br#[J]. 吉林化工学院学报, 2025, 42(11): 90-96.
ZHANG Jun, WU Haocheng, HAN Bo, YOU Tao, WEI ZiXuan, FENG Xin, XIN Ruihao. DAIF-Cox:A Multimodal Fusion-Based Model for Breast Cancer Prognostic Risk Prediction#br#. Journal of Jilin Institute of Chemical Technology, 2025, 42(11): 90-96.
