Supervised learning system for detection of cardiac arrhythmias based on electrocardiographic data

• Published in: 2019 IEEE International Conference on E-health Networking, Application & Services (HealthCom) pp. 1 - 4
• Main Authors: Castillo, Oscar Daniel Diaz, Ayala, Luz Janneth Parra, Perez, Airan Leonardo Carreno, Jimenez, Jose Misael Cubillos, Castano, Yohana Lyceth Corba, Munoz-Sarmiento, Diana Marcela
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-10-2019
• Subjects: Biological neural networks; Cardiac arrhythmia; Deep learning; Diseases; ECG; Electrocardiography; Heart; Machine learning; Neuronal networks; Training
• DOI: 10.1109/HealthCom46333.2019.9009601

Heart diseases are one of the leading causes of death around the world, and it is a mayor health issue in many countries. Among the large number of heart diseases, cardiac arrhythmias are a group of conditions where the heartbeat is irregular, and it can predispose a person to have a stroke or heart failure. Doctors use the electrocardiogram (ECG) taken from the patient to diagnose the presence of a particular arrhythmia condition, usually by means of manually analyzing and interpreting the comprising waves.This paper presents a proposal to design a tool for detection of cardiac arrhythmias based on ECG data, using supervised learning techniques found on the Python machine learning library called PyTorch, and using publicly available processed ECG parameters from the UCI Arrhythmia Data Set as input for training and verification of our neural network based model. Tool detection accuracy using 450 samples for training achieved around 85 % of correct answers.Additionally, this tool was designed as an assistant tool for aiding doctors to better diagnose these kind of conditions, taking into account a ease-to-use interface. Finally the paper discuss the benefits and potential of machine learning techniques and artificial intelligence applied to assist medical diagnosis processes.