Towards a Pre-Processing Algorithm for Automated Arrhythmia Detection

There are a variety of different wearable fitness/cardiac monitoring devices that are currently used in many people’s day to day life. The primary cardiac function of these devices is to monitor heart rate, however we believe that they could be utilized to detect different forms of arrhythmia. In order to categorize and identify different forms of arrhythmia, we are utilizing published EKG data sets from existing databases as a basis for machine learning. The challenge that comes from the existing data sets is that the format they present the data in does not lend itself to machine learning, which requires data to be in a vector. This makes the process of converting the existing data sets into workable vectors long and tedious. Therefore, we are working to develop an algorithm that will be able to vectorize the data from multiple different data sets so we, and anyone who wishes to use machine learning on these signals, are able to quickly and accurately use now workable, prior data sets.https://ecommons.udayton.edu/stander_posters/2605/thumbnail.jp