Summary Cancer remains the leading cause of death from disease in children. Development of therapeutic options for the remaining lethal cancers has seen little progress, hampered by the rarity of childhood cancers and institutionally isolated data systems holding tumor biomarker, genetic, genomic, treatment and clinical data, which impedes maximally powered therapeutic studies. The National Cancer Institute’s Childhood Cancer Data Initiative (CCDI) seeks innovation in pediatric cancer research approaches by markedly increasing data-sharing. Under the auspices of a previous P30 Supplement award, the USC Norris Comprehensive Cancer Center (NCCC) in partnership with Children’s Hospital Los Angeles (CHLA) successfully curated and contributed to CCDI genomic and clinical data of 1039 patient of three major categories (hematopoietic malignancies, solid tumors and CNS tumors) and 186 subtypes. We now propose to enrich the data sets that we submitted and to develop an online diagnostic resource for pediatric cancers driven by augmented Artificial Intelligence (A2I), which aims to improve pediatric cancer care access and affordability by providing a scalable and standardized diagnostic process. The proposed A2I system will develop an AI-powered classifier for pediatric CNS and sarcomas, and ultimately all pediatric cancer, using whole-slide images and molecular findings in combination. Aim 1 will collect whole-slide image (WSI) from 599 solid tumors and whole-genome methylome data of 200 CNS tumors. Collected WSI and methylation data of these 599 tumors will be contributed to CCDI and become an integral part of our existing CHLA CCDI data set. Aim 2 will develop a multi-modal classifier of sarcomas and CNS tumors using an Augmented AI (A2I) framework. The proposed classifier is entitled Multi-Modal AI-based Diagnosis for Pediatric Oncology (MAD4PO), which will be cloud-based and web-accessible. To build this classifier, we will leverage the Amazon Web Services (AWS) A2I framework and associated services and tools to facilitate human-AI collaboration for optimal diagnostics, and to scale out access to the developed ML/AI models for global healthcare providers. Work carried out under this supplement will facilitate efforts to understand the biologic basis of childhood cancers and to develop improved treatment for these diseases, while providing new tools for more rapid and accurate diagnosis of pediatric cancers.