APPARATUS FOR HEART DISEASE CLASSIFICATION BASED ON DEEP LEARNING, AND METHOD THEREFOR

摘要

The present invention relates to an apparatus for diagnosing heart disease using a deep learning-based electrocardiogram, and a method therefor. The apparatus for diagnosing heart disease using a deep learning-based electrocardiogram according to the present invention comprises: a data input unit which receives an input of normal electrocardiogram data and electrocardiogram data diagnosed as arrhythmia; a data classification unit which classifies the input arrhythmia data into data of a plurality of disease symptoms; a training unit which generates a dataset by randomly extracting data from the normal electrocardiogram data and the arrhythmia data classified into a plurality categories, and inputs the generated dataset into a plurality of deep learning models to train the models on pathological characteristics of arrhythmia; and a control unit which receives an input of electrocardiogram data of a subject to be diagnosed for disease symptoms, and performs a diagnosis by inputting the input electrocardiogram data into a pre-established classification model in stages so as to classify same into highly accurate disease symptoms. According to the present invention, diagnosis results are not classified into binary categories, such as normal and abnormal (arrhythmia), but can instead be classified into multiple categories, such as normal, arrhythmia type 1, arrhythmia type 2, arrhythmia type 3, … N, and thus, accurate diagnosis results can be provided.