Abstract

The progression of multiple myeloma (MM) from precursor conditions, such as monoclonal gammopathy of undetermined significance (MGUS), is driven by a complex network of genomic and molecular alterations. This thesis integrates traditional bioinformatics with advanced artificial intelligence (AI) and machine learning (ML) models to identify key genomic drivers of disease progression. Specifically, deep learning is applied to the publicly available GSE6477 dataset using generative adversarial networks (GANs), artificial neural networks (ANNs), support vector machines (SVMs), and XGBoost, alongside statistical tools like GEO2R and platforms such as STRING, Cytoscape, Enrichr, and KMplotter. These methods help identify gene signatures involved in the transition from a healthy state to MGUS and from MGUS to MM. Reverse-engineering gene expression networks and applying feature selection techniques, including principal component analysis (PCA) and random forests, enabled the identification of significant genes at each stage of disease progression. To address challenges such as class imbalance and data scarcity, synthetic data generation using GANs improved the robustness of prognostic models by generating realistic samples, validated through Euclidean distance and Pearson correlation to ensure biological fidelity. Key findings include identifying gene networks strongly associated with MM progression, validated through enrichment and survival analyses using KMplotter. AI/ML models, particularly those enhanced by synthetic data, showed improved generalizability across patient cohorts, suggesting potential broader applications beyond the GSE6477 dataset. This research provides insights into the molecular mechanisms underlying MM progression, which may aid in developing targeted therapies. Moreover, combining traditional bioinformatics with ML techniques demonstrates a powerful approach for deciphering gene interactions in complex diseases like MM. Future work should focus on translating these findings into clinical practice to enhance patient outcomes through personalized medicine