Key result
GWO-optimized CNN-HRV fusion improves ECG abnormality detection accuracy by ~7 percentage points over Genetic Algorithm.
Why the study?
ECG-based cardiac abnormality risk detection remains challenging under limited-data and cross-patient conditions, motivating the development of a metaheuristic-optimized CNN-HRV fusion framework combining raw ECG windows and HRV features.
Does a Grey Wolf Optimization-optimized CNN-HRV fusion framework improve ECG-based cardiac abnormality risk detection accuracy and F1-score compared to Genetic Algorithm optimization in limited cross-patient ECG data from older adults?
Does a Grey Wolf Optimization-optimized CNN-HRV fusion framework improve ECG-based cardiac abnormality risk detection accuracy and F1-score compared to Genetic Algorithm optimization in limited cross-patient ECG data from older adults?
A Grey Wolf Optimization-tuned CNN-HRV fusion model demonstrates high accuracy and noise robustness for early computational ECG abnormality risk screening under severe cross-patient data constraints, though clinical validation with physician-annotated datasets is still required.
GWO-optimized CNN-HRV fusion may enhance ECG detection in small datasets; hypothesis-generating and should not yet change practice.
ECG-based cardiac abnormality risk detection remains challenging under limited-data and cross-patient conditions. This study proposes a metaheuristic-optimized CNN-HRV fusion framework that combines raw ECG windows and HRV features. From 27 ECG recordings of 8 patients, 372 windows were generated and labeled through K-Means-assisted labeling into 280 normal-risk and 92 abnormal-risk samples. In valid cross-patient scenarios, the proposed framework achieved an accuracy of 0.913413, a sensitivity of 0.935185, an F1-score of 0.875214, and a ROC-AUC of 0.985714. Across 10 runs, Grey Wolf Optimization (GWO) outperformed Genetic Algorithm (GA) in accuracy (0.912533 vs. 0.845016, p = 0.0149228) and F1-score (0.840315 vs. 0.664700, p = 0.0278711). These findings indicate that the GWO-optimized CNN-HRV fusion is promising for ECG abnormality screening, while the pseudo-labeling strategy positions the method as early risk stratification rather than final clinical diagnosis.
No takes yet. Share an insight, caveat, or question.
Prenata et al. (2026) studied ECG-based cardiac abnormality risk (n=8). Grey Wolf Optimization (GWO)-optimized CNN-HRV fusion framework vs. Genetic Algorithm (GA)-optimized CNN-HRV fusion framework was evaluated on Mean classification accuracy across 10 runs for cardiac abnormality risk detection (p=0.0149228). GWO-optimized CNN-HRV fusion significantly improved ECG abnormality detection accuracy (0.9125 vs. 0.8450, p=0.0149) and F1-score (0.8403 vs. 0.6647, p=0.0279) compared to Genetic Algorithm.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: