Abstract

Pediatric Acute Lymphoblastic Leukemia (ALL) is the most common childhood cancer, and due to its heterogeneity, careful diagnosis and treatment are required. AI and ML stand at the frontiers of clinical outcomes by positively affecting diagnosis, risk stratification, and chemotherapy selection. This systematic review assesses the role of AI and ML in the diagnosis and treatment of pediatric ALL, specifically focusing on diagnostic accuracy, personalized treatment, and overcoming challenges such as drug resistance and data limitations. The review adhered to PRISMA 2020 guidelines and carried out literature searches on Google Scholar, Web of Science, and PubMed using a focused search string. Of 979 initially screened articles, 50 were enlisted for review based on AI applications in the diagnosis and treatment of pediatric ALL. Data extraction captured AI methodologies, performance metrics, and clinical outcomes of interest. AI-based models demonstrated better accuracy in diagnosis than traditional techniques. Deep learning models (including CNN, transformers) were shown to outperform traditional means for various applications such as flow cytometry-based identification of minimal residual disease, genomic biomarker analyses, and bone marrow biopsy interpretations. Optimization of chemotherapy using AI increased survival through dose selection and minimized toxicity through prediction of adverse events. Federated learning and explainable AI (XAI) became imperative paradigms enabling privacy preservation, data heterogeneity, and fostering clinician trust. The recommendations stress the need for the integration of AI-led predictive modeling, federated learning, and multi-institutional validation to improve clinical workflows, personalize treatment, and improve pediatric Acute Lymphoblastic Leukemia (ALL) outcomes.