Table 7 Statistical comparison of CLCD-Block vs. existing models.
From: Blockchain enabled collective and combined deep learning framework for COVID19 diagnosis
Comparison model | p-value (Wilcoxon Test) | Significance (p < 0.05) | Implication |
|---|---|---|---|
EfficientNetB425 | \(2.38 \times 10^{-9}\) | Significant | CLCD-Block significantly outperforms in all metrics |
Deep CNN31 | \(1.647 \times 10^{-8}\) | Significant | Consistently better results across datasets |
Blockchain-based FL38 | \(7.372 \times 10^{-8}\) | Significant | Superior robustness and diagnostic precision |
XGBoost26 | \(1.4497 \times 10^{-7}\) | Significant | Outperforms especially in recall and specificity |
CAP-CNN41 | \(1.54972 \times 10^{-6}\) | Significant | More generalizable and accurate detection |
LSTM + Blockchain14 | \(2.000163 \times 10^{-5}\) | Significant | Better recall and overall balance in performance |
Bi-LSTM + Blockchain33 | \(1.42 \times 10^{-4}\) | Significant | Marginal but significant improvement in F1 and accuracy |
