DWPL-GCNMF: Structure-Aware Dynamic Weighted Pseudo-Label Learning for Adverse Drug Reaction Prediction
Clinical Snapshot
PICO Framework
| P — Population | Drug-ADR association datasets: DrugBank (1,177 drugs × 4,247 ADRs) and SIDER (1,080 drugs × 5,579 ADRs) |
| I — Intervention | DWPL-GCNMF framework combining graph convolutional networks with matrix factorization and dynamic weighted pseudo-label learning |
| C — Comparator | Representative baseline machine learning methods for ADR prediction |
| O — Outcomes | AUPR, F1 scores, and top-K recall metrics for ADR prediction accuracy |
Bottom Line
This computational study presents DWPL-GCNMF, a novel machine learning framework for predicting adverse drug reactions that combines graph neural networks with dynamic pseudo-label learning. The method shows promising performance on standard benchmarks, achieving AUPR scores of 0.67 and 0.61 on DrugBank and SIDER datasets respectively. While the technical innovation addresses important challenges in ADR prediction including data sparsity and confirmation bias, the clinical utility remains unproven. The lack of statistical significance testing, confidence intervals, and real-world validation limits immediate clinical application. For Australian practice, this represents an interesting research direction that could eventually support TGA pharmacovigilance efforts, but substantial additional validation would be required before implementation in clinical decision-making or regulatory processes.
Key Findings
P Value: Not reported
Effect Size: AUPR: 0.6656 (DrugBank), 0.6106 (SIDER); F1: 0.6194 (DrugBank), 0.5816 (SIDER)
Primary Outcome: ADR prediction accuracy measured by AUPR and F1 scores
Nnt Or Sensitivity: Recall@15: 0.8078 (DrugBank), 0.8146 (SIDER)
Confidence Interval: Not reported
Clinical Application
Requires computational infrastructure and integration with existing drug safety systems Could support TGA post-market surveillance activities and complement ADRAC reporting, but requires validation with Australian drug safety data Potentially applicable to drug safety surveillance and pharmacovigilance programs
Abstract
Adverse drug reaction (ADR) prediction is typically formulated as drug-ADR association learning on extremely sparse, positive-unlabeled matrices, which limits generalization and makes naive pseudolabel expansion prone to confirmation bias. Here, we propose DWPL-GCNMF, a structure-aware semisupervised framework that combines graph convolutional network (GCN)-based structural embeddings learned from a drug-protein knowledge graph with matrix factorization scoring and introduces a prediction-consistency-driven dynamic weighted pseudolabel learning strategy to expand training pairs while down-weighting unreliable pseudolabels; multiple base models are further fused to improve prediction stability. We evaluate DWPL-GCNMF on a DrugBank-derived benchmark (1,177 drugs × 4,247 ADRs; ∼2.6% observed associations) and an external SIDER benchmark (1,080 drugs × 5,579 ADRs; ∼2.3% observed associations) to assess robustness under different data sets and increased sparsity. DWPL-GCNMF consistently outperforms representative baselines, achieving AUPR/F1 of 0.6656/0.6194 on DrugBank and 0.6106/0.5816 on SIDER, while also improving top-K prioritization (e.g., Recall@15 up to 0.8078 and 0.8146, respectively), indicating that structure-aware representation learning with dynamically weighted pseudolabeling provides a robust approach for prioritizing potential ADRs under severe data sparsity.
References
- 1.Chen, H., & Luo, K. (2026). DWPL-GCNMF: Structure-Aware Dynamic Weighted Pseudo-Label Learning for Adverse Drug Reaction Prediction. Journal of Chemical Information and Modeling. https://doi.org/10.1021/acs.jcim.6c00294
Related Research
CPT: pharmacometrics & systems pharmacology
Structured Schemas for Provenance-Rich, LLM-Assisted QSP Model Calibration
2 Aug 2026
Journal of pharmacokinetics and pharmacodynamics
Diffusion models for virtual populations and pharmacometric simulations
1 Aug 2026
Journal of chemical information and modeling
HyperDC: A Non-Uniform Hypergraph Framework for Dual- and Higher-Order Drug Combination Recommendation Across Diverse Complex Diseases
29 July 2026
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service