The advent of machine and deep learning algorithms has forever changed the way academic and industry researchers approach real-world problems that come with big data. With volume and variety of data being surpassed only by the increasing need to make sense of it, novel tools and approaches suited for handling increasingly complex data need to be explored. Topology has recently seen progress in this direction. The idea of studying algebro-topological structures from point-cloud represented data has shown tremendous potential in extracting useful information about the underlying data set. Applications abound in varied fields including neuroscience, molecular biology, and even social network analysis. In this talk, we present a novel approach to the 2017 Physionet/Computing in Cardiology Challenge of classifying single-lead electrocardiograms to diagnose Atrial Fibrillation by integrating topology-based features to train a simple machine learning algorithm. Via delay embeddings, we transform time series data into high dimensional point-clouds to convert periodic information to algebraically computable topological signatures. We show how feeding statistical summaries of these features to a random forest model provides a significant increase on classification accuracy when compared to standard features used by medical professionals to diagnose Atrial Fibrillation. The talk will be accessible to a general audience with no prior background on Topology or Machine Learning. This is joint work with Christopher Dunstan (U. Maryland Baltimore County), Esteban Escobar (Cal Poly State U.), Luke Trujillo (Harvey Mudd), and Dr. David Uminsky (U. San Francisco).