Hybrid AI Framework for Multi-Omics-based Kidney Tumor Subtype Classification and Precision Oncology

Abstract

This study presents an AI-driven multi-omics framework designed to improve the classification of kidney tumor subtypes and support personalized treatment strategies in precision oncology. By integrating genomic, transcriptomic, and proteomic data from publicly available TCGA repositories, the proposed model builds comprehensive molecular profiles of kidney tumors. The system achieves a classification accuracy exceeding 96% by combining machine learning and deep learning techniques — specifically Random Forest (RF) for feature selection, Support Vector Machines (SVM) for handling high-dimensional data, Convolutional Neural Networks (CNNs) for spatial pattern extraction, and Transformer models to capture contextual relationships across biologically ordered gene sequences. Unlike conventional ensemble approaches, this hybrid framework is optimized for both predictive accuracy and computational efficiency, making it suitable for real-time clinical use. Additionally, the model identifies key tumor-specific biomarkers that can guide individualized therapy. Experimental results confirm that the proposed system significantly outperforms traditional diagnostic methods and single-omics models. Designed for integration with cloud-based Clinical Decision Support Systems (CDSS), the framework has strong potential to enhance real-time oncology decision-making.