Pediatric hematological pathologies, including acute myeloid leukemia (AML) and thrombocytopenia absent radius (TAR) syndrome, are complex blood disorders with significant morbidity and mortality, yet their molecular mechanisms remain incompletely understood. This dissertation investigates the role of alternative splicing in these diseases in parallel with the development and application of advanced computational tools for large-scale RNA-sequencing analysis. Using rMATS-turbo, a software capable of efficiently detecting and quantifying splicing events across thousands of samples, we identified disease-associated splicing events across diverse conditions (e.g., diseased vs. healthy tissue). We extended this functionality to cloud-based platforms via rMATS-cloud, enhancing scalability and enabling collaborative data analysis. To address biases in splicing analysis due to variable gene expression levels, we developed rMATS-DARTS, which employs a Bayesian framework to improve detection accuracy, especially in datasets with low read depth. Applying these tools to TAR syndrome revealed specific splicing events associated with RBM8A mutations that disrupt megakaryocyte maturation, potentially contributing to the disease's clinical manifestations. In pediatric AML, we identified high-risk splicing patterns and potential predictive biomarkers linked to cytogenetic abnormalities and mutation status, providing insights for risk stratification and personalized treatment approaches. These findings underscore the importance of alternative splicing in understanding disease progression and highlight its potential as a target for therapeutic intervention. In conclusion, this dissertation demonstrates that when optimized through computational advancements, alternative splicing analysis significantly enhances our understanding of pediatric disease. This research provides a foundation for integrating splicing analysis into clinical frameworks, advancing diagnostic accuracy, and improving therapeutic strategies.