Identifying and characterizing cis-regulatory RNAs within mRNAs, key modulators of stability and translation, represents a significant gap in human genome annotation. These RNAs, reliant on precise RNA structures, group into related families known as structural RNA families. We have developed the EvoFam genomics model, a novel tool for capturing the evolutionary mutation patterns of these conserved RNA structures, thus facilitating a genome-wide identification of human RNA structure families. Building on this groundwork, this project advances the development of innovative machine learning models designed to unearth conserved cis-regulatory RNA structures across vertebrate genomes. This initiative includes the creation of new RNA structure 2D/3D modeling and comparative genomics methods, underpinned by robust neural network architectures. Such advancements will increase the sensitivity of our methods for detecting structural RNA families, resulting in a significantly expanded catalog of these families. The culmination of this research promises computational breakthroughs in RNA modeling and opens up translational opportunities, including the targeted use of structural RNA-based therapeutics. This project is a collaboration with Brian Parker at School of Computing, ANU.