A robust deep convolutional neural network with batch-weighted loss for heartbeat classification

作者：

Highlights：

• Classified imbalanced ECG heartbeats using a novel batch-weighted loss formula.

• Trained a robust classifier with raw 1-lead ECG data without any preprocessing.

• Achieved high classification performance under intra/inter-patient paradigms.

• Improved classification performance using multiple heartbeats as input.

• Our method is easily applicable to time series datasets from other domains.

摘要

•Classified imbalanced ECG heartbeats using a novel batch-weighted loss formula.•Trained a robust classifier with raw 1-lead ECG data without any preprocessing.•Achieved high classification performance under intra/inter-patient paradigms.•Improved classification performance using multiple heartbeats as input.•Our method is easily applicable to time series datasets from other domains.

论文关键词：

论文评审过程：Received 14 June 2018, Revised 24 November 2018, Accepted 19 December 2018, Available online 21 December 2018, Version of Record 3 January 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.12.037