An uncertainty reasoning method for abnormal ECG detection

Abstract

The electrocardiogram (ECG) recognition is important for cardiovascular disease monitoring. It is significant to investigate automatic diagnosis methods related to wearable ECG instruments. This paper introduces Certainty Factor model based an uncertainty reasoning method for abnormal detection. It discusses the application and improvement of Certainty Factor model based on experts' experience in electrocardiogram diagnosis and puts forward the thought of determining the model parameters by machine learning. The experiment results show that the improved Certainty Factor model has better accuracy. The stability of Certainty Factor model is better than that of Bayes when the number of the disease type is increased.