The goal of this project is to discover knowledge and patterns in real-world 'big' and 'small' (limited) data samples by engineering data-driven models such as machine and deep-learning algorithms. Real-world datasets come with a number of challenges that are not present in publicly available 'vanilla' datasets used to evaluate new machine learning architecture and/or algorithms in the literature. A model that performs excellently on the MNIST data set may not reproduce similar performance on a more complex real-world data set. This project, in collaboration with a number of medical schools, aims to leverage electronic health records data to 1) validate newly proposed neural network architecture, 2) to introduce artificial intelligence to solve medical research questions and bridge the gaps between the technical and clinical worlds, 3) to develop cross-disciplinary data science education and research capacity at the home institution.