AI based multiomics integration for cancer diagnosis and prognosis

Cancer remains a critical global health challenge, driving the need for innovative approaches in diagnosis and treatment. This research introduces OmicsFusionNet, an AI-powered hybrid model integrating machine learning and deep learning to revolutionize cancer care. The tool incorporates up to six multiomics datasets-genomics, transcriptomics, and epigenomics-achieving 80.2% accuracy across 23 cancer types. Notably, RNAseq and methylation integration reached 99.8% accuracy, highlighting XGBoost's feature selection and deep learning's classification strength. For ovarian cancer stage detection, OmicsFusionNet optimized analysis using CPTAC-OV and TCGA-OV datasets, achieving accuracies between 83% and 91% by combining ElasticNet and XGBoost with deep learning. Additionally, KEGG pathway enrichment of multiomics biomarkers identified key cancer-related pathways, advancing early detection, biomarker discovery, and personalized treatments. This study underscores the transformative potential of AI and multiomics integration in cancer research, enabling precise interventions and uncovering novel mechanisms that enhance patient outcomes.