Automated detection of cardiovascular disease by electrocardiogram signal analysis: a deep learning system
Problem
Manual ECG interpretation is error-prone; automated detection and classification of cardiovascular disease from ECG signals was needed.
Why it matters
Misdiagnosis and missed diagnosis in primary care affect patient outcomes and hospital manpower costs.
What the AI did
A CNN was trained on 259,789 ECG signals to classify ECGs into 18 classes (17 abnormalities plus normal) and validated on an independent test set.
Results
Overall accuracy close to 95%; 99.15% accuracy for normal rhythm/atrial fibrillation on an independent test set of 18,018 ECGs.
Impact
Authors claim it could reduce misdiagnosis in primary care and save manpower in large hospitals; deployment not shown.
Record
| Venue | Cardiovascular Diagnosis and Therapy |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Xin Zhang, Kai Gu, Shumei Miao, Xiaoliang Zhang, Yuechuchu Yin, Cheng Wan, Yun Yu, Jie Hu et al. |
| Citations | 81 (OpenAlex, at import time) |
| References | 24 |
| Model or method | Convolutional neural network (CNN) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.21037/cdt.2019.12.10 |
| Open access | diamond |
| Keywords | ECG, deep learning, CNN, cardiovascular disease, automated diagnosis |
Importance rating
significant — thousands to millions affected
Automated ECG diagnosis covering >90% of clinical diagnoses could reduce misdiagnosis/missed diagnosis in primary care and save hospital manpower; abstract gives no scale of affected patients.