Deep-Learning Based Heartbeat Analysis for the Detection of Cardiac Arrhythmia
By Shu Hua (Jenny) Li Intermediate Category (Grades 9-10) Innovation | Big Data / AI, Biology, Engineering and Computer Science
Cardiac arrhythmia is a prevalent heart condition where a patient’s heart beats irregularly due to improperly functioning electrical impulses. Though arrhythmias are often treatable, they can be difficult to diagnose and dangerous if left undetected, leading to strokes and even heart failure. In this project, a convolutional neural network that analyzes an ECG recording of a heartbeat for the classification of healthy cases and four types of cardiac arrhythmia has been created. The final model had varying accuracies for each of the diseases, with 97% of unhealthy cases classified as unhealthy and 98% of healthy cases classified as healthy, therefore putting it a step above similar existing technologies where cardiologists must review all results.
