Ha Tang Vinh

Accurately predicting the molecular subtype of cancer is crucial for personalized diagnosis and treatment. Furthermore, finding reliable biomarkers is essential for achieving personalized medicine and improving patient outcomes overall. This paper presents a novel approach for multi-omics biomarker discovery in Glioblastoma using explainable artificial intelligence. The proposed method integrated data from the TCGA cohort and MOGONET's deep learning model for subtyping problems and then used the Integrated Gradients algorithm to discover important features in the multi-omics data. The identified biomarkers were subsequently validated through two approaches: first, by comparing them with established ground truth data from reference databases, and second, performance evaluation using classical machine learning models.

The integration of multi-omics data has advanced cancer subtype classification, yet the deep learning models employed often lack interpretability. Integrated Gradients (IG), an explainable AI (XAI) method, addresses this by attributing model predictions to input features, thereby facilitating biomarker discovery. While prior studies typically apply IG to a single architecture, this research systematically benchmarks IG across three distinct deep learning frameworks: MOGONET, MMDynamics, and HTML, using multi-omics data from The Cancer Genome Atlas (TCGA). We leverage IG to identify and compare top-ranked biomarkers, assessing their biological relevance and reproducibility across models. Our experiments, conducted with stratified K-fold cross-validation. This work provides practical guidance on selecting appropriate deep learning architectures and IG baselines for multi-omics research, aiming to foster more interpretable and clinically meaningful applications of AI in oncology.