Table 6 Comparative analysis of the proposed model.
From: A fused weighted federated learning-based adaptive approach for early-stage drug prediction
Model | Type | Accuracy (%) | Miss rate (%) | Highlights |
|---|---|---|---|---|
ANN + SVM26 | Traditional ML | 73.1–76.0 | 24–26.9 | Baseline comparison |
SVM, LR, NB27 | Classical ML | 69–80 | 20–31 | Limited to centralized settings |
DeepDrug | Deep learning | 89.4 | 10.6 | Deep graph embeddings |
ChemBERTa | Transformer | 91.1 | 8.9 | BERT-based SMILES representation |
FL-Mol | Federated learning | 90.2 | 9.8 | FL for molecular property prediction |
FedHealthNet | Personalized FL | 91.5 | 8.5 | Personalized layers per client |
Proposed FWAFL | Adaptive FL + Fusion | 91.9 | 8.1 | Client-weighted fusion + adaptivity |
