Multi-Omics Analysis for Improved Diagnosis, Prognosis, and Drug Response Prediction in Digestive System Tumors

Lin Zhou,Zhengzhi Zhu,Xinmiao Wang,Hongbo Gao
DOI: https://doi.org/10.1109/ICUS58632.2023.10318469
2023-01-01
Abstract:Digestive system tumors (DST) have been the leading cause of cancer-related deaths globally in recent decades, making accurate tumor diagnosis, prognosis improvement, and drug response prediction crucial issues in personalized medicine. With the explosion in massive biological data and the transition from traditional biostatistical methods to computer-assisted approaches, AI-based diagnostic support systems have shown great potential in medical computer-assisted fields. In this study, we propose a novel multi-omics integration approach, Multi-Omics Fusion Graph Convolutional Network (MFGCN), to construct diagnostic models, survival prediction models, and drug response prediction models for DST. Firstly, we calculate the similarity matrix of each omics data and use Graph Transformer (GT) to learn the new graph structure. Secondly, we use GCN to learn features of each omics data. Finally, to integrate features of different omics for prediction, we use View Correlation Discovery Network (VCDN). Our experimental results outperform other currently advanced methods, offering significant potential value for personalized medicine of DST.
What problem does this paper attempt to address?