Abstrait

Deep learning-based artificial intelligence for predicting risk and prognosis in patients with cardiovascular disease

Ki-Hyun Jeon, Joon-myoung Kwon, Kyung-Hee Kim, Jinsik Park

Cardiovascular disease (CVD) is a major healthcare problem worldwide. Risk stratification and prognosis prediction are critical in identifying high-risk patients and in decision making to devise treatment strategies for patients with CVD. For this purpose, various models have been developed and validated against large amounts of population registry data by using conventional statistical methods such as regression-based models. However, these conventional models have a problem of over-generalization and are not applicable to all individual patients. Deep learning is a branch of artificial intelligence in which artificial neural networks are used to analyze data patterns; it is similar to functioning of the human neural system. An advantage of deep learning is the automatic learning of features and relationships from given data. Recently, deep learning achieved high performance in several medical domains, such as image classification, diagnosis, clinical outcome prediction, and gene analysis. The focus of this review is to summarize deep learning-based prediction models in patients with CVD in terms of accuracy in comparison with conventional models.