Explaining deep neural networks for knowledge discovery in electrocardiogram analysis
importance 4/10
ai assisted
ECG Monitoring and Analysis cardiology stage: research
Problem
Deep learning ECG predictions lack human-understandable explanations, limiting clinical trust and knowledge discovery.
Why it matters
Doctors remain responsible for decisions and need to understand AI predictions before using them.
What the AI did
ECGradCAM generates attention maps that reveal how a deep learning model measures ECG amplitudes and intervals, and can help develop novel ECG features.
Results
not stated in source
Impact
Qualitative demonstration of explainable ECG analysis; no quantitative benchmark in abstract.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Steven Alexander Hicks, Jonas Lynggaard Isaksen, Vajira Lasantha Thambawita, Jonas Ghouse, Gustav Ahlberg, Allan R Linneberg, Niels Grarup, Inga Strümke et al. |
| Citations | 75 (OpenAlex, at import time) |
| References | 45 |
| Model or method | ECGradCAM with a deep neural network (base model not named) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-021-90285-5 |
| Open access | gold |
| Keywords | ECG, deep learning, explainability, ECGradCAM, attention maps |
Importance rating
4
real but narrower advance
Abstract states doctors are responsible for decisions and need understandable explanations; a methodological advance for ECG AI trust and knowledge discovery, narrow scope, no scale given.