Boyang Huang
Pronouns: he/him/his
Research Mentor(s): Mohammed Saeed, Faculty Member / Clinical Lecturer
Research Mentor School/College/Department: Department of Internal Medicine, Michigan Medicine
Presentation Date: Thursday, April 22, 2021
Session: Session 4 (2pm-2:50pm)
Breakout Room: Room 17
Presenter: 7
Abstract
Atrial fibrillation is the most common type of heart arrythmia in humans, and our goal is to develop a method of characterizing medical data signals to assist cardiologists to better manage atrial fibrillation. Predicting atrial fibrillation from ECG signals because ECG patterns may change from patient to patient, and include noise. In our study, we provide a comparison across different existing methods of classifying ECG time series using machine learning models. We tested two different machine learning models: a Fully Connected Neural Network and a Support Vector Machine, and compared their performance and accuracy on the Physionet Atrial Fibrillation Challenge Dataset (www.physionet.org). We found that although a Fully Connected Neural Network as a deep learning model is pretty robust when given large amounts of data, the Support Vector Machine method performed better than deep learning when limited data is available to use. Potential applications of this include live analysis of ECG signals that could assist cardiologists to perform diagnosis in real time.
Authors: Mohammed Saeed, Boyang Huang
Research Method: Clinical Research
