What is the clinical evidence from this study?

Study design: Other. Population: Atrial fibrillation (n=155). Intervention: Resnet deep learning models vs. External datasets. Primary outcome: Accuracy of AF detection.

September 14, 2021Open Access

ECG data dependency for atrial fibrillation detection based on residual networks

Key Result

Resnet-based atrial fibrillation detection models achieved 98-99% accuracy on internal datasets but dropped to 53-92% on external datasets, demonstrating significant data dependency.

Structured PICO

Does the use of residual neural networks for AF detection show data dependency when evaluated on external datasets?

Population

ECG signals from three different databases from PhysioNet

Intervention

Deep learning-based AF detection algorithm using residual neural networks (Resnet 18, 34, 50, and 152)

Comparator

Evaluation on external datasets versus internal datasets

Outcome

Accuracy of AF detectionsurrogate

Deep learning models for AF detection exhibit significant data dependency, performing worse on external datasets, though this effect is mitigated by larger training datasets.

Limitations

Implemented with only one deep learning architecture (Resnet)
Non-AF class in AFDB is composed of all other rhythms due to absence of normal sinus rhythm annotation
Data used is from only three open-source databases

Abstract

Atrial fibrillation (AF) is an arrhythmia that can cause blood clot and may lead to stroke and heart failure. To detect AF, deep learning-based detection algorithms have recently been developed. However, deep learning models were often trained with limited datasets and were evaluated within the same datasets, which makes their performance generally drops on the external datasets, known as data dependency. For this study, three different databases from PhysioNet were used to investigate the data dependency of deep learning-based AF detection algorithm using the residual neural network (Resnet). Resnet 18, 34, 50 and 152 model were trained with raw electrocardiogram (ECG) signal extracted from independent database. The highest accuracy was about 98-99% which is evaluation results of test dataset from the own database. On the other hand, the lowest accuracy was about 53-92% which was evaluation results of the external dataset extracted from different source. There are data dependency according to the train dataset and the test dataset. However, the data dependency decreased as a large amount of train data.

AIに質問

Bookmark

View Full Paper