A Disentangled VAE-BiLSTM Model for Heart Rate Anomaly Detection

• DOI: 10.3390/bioengineering10060683
• Publication Date: 2023-06-05T06:57:47Z
• Authors: Alessio Staffini; Thomas Svensson; Ung-il Chung; Akiko Kishi Svensson
• Keywords: Anomaly (physics)

Cardiovascular diseases (CVDs) remain a leading cause of death globally. According to the American Heart Association, approximately 19.1 million deaths were attributed CVDs in 2020, particular, ischemic heart disease and stroke. Several known risk factors for include smoking, alcohol consumption, lack regular physical activity, diabetes. The last decade has been characterized by widespread diffusion use wristband-style wearable devices which can monitor collect rate data, among other information. Wearable allow analysis interpretation physiological activity data obtained from wearer therefore be used prevent potential CVDs. However, these are often provided manner that does not general user immediately comprehend possible health risks, require further analytics draw meaningful conclusions. In this paper, we propose disentangled variational autoencoder (