Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance
Abstract
IntroductionAdverse drug reaction (ADR) prediction remains challenging because spontaneous reporting data leave many exposure unit–ADR associations unobserved and provide limited support for candidate prioritization and mechanistic interpretation.MethodsWe developed MEG-HHGN, a mechanistic evidence-guided hyperbolic heterogeneous graph neural network. It represents exposure units as combinations of standardized drug names and routes of administration and integrates spontaneous reports with compounds, drug targets, ADR-related targets, and protein–protein interactions. Using 73,530 spontaneous reports collected in Jiangsu Province from 2010 to 2023, we evaluated MEG-HHGN against eight baseline methods through five-fold cross-validation.ResultsMEG-HHGN achieved AUC/AP/MCC values of 0.9550/0.9537/0.7917 in known-exposure-unit completion and 0.9090/0.9064/0.6587 in side-information-supported exposure-unit cold-start evaluation. Global scoring of unobserved exposure unit–ADR pairs produced a ranked candidate pool for pharmacovigilance review. Representative cases involving intravenous vinorelbine tartrate–myelosuppression and oral aspirin–proteinuria illustrated PPI-bridged and shared-target pathways, respectively.DiscussionMEG-HHGN integrates route-aware prediction, candidate prioritization, and pathway-based interpretation in a unified framework to support signal triage, mechanistic hypothesis generation, and evidence review in post-marketing pharmacovigilance.
Related
- Guidance on the interpretation of pharmacovigilance signal detection data and analyses
- AlphaGenome Atlas: in silico mutagenesis of the entire human genome improves prioritization and interpretation of non-coding variants
- Transparent Weighting of Heterogeneous Evidence for Auditable Candidate Prioritization
- Integrating pharmacovigilance and machine learning: bleeding signal detection and risk prediction for oral anticoagulants in cancer patients
- Machine learning prediction model for distant metastasis at diagnosis in pediatric neuroblastoma: development, interpretation, and temporal validation
BibTeX
@article{lv2026hyperbolic,
title = {{Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance}},
author = {Ru Lv and Xiaoyu Xu and Yulin Ji and Jianxiang Wei and Zuojian Zhou and Kankan She},
journal = {Frontiers in Pharmacology},
year = {2026},
month = nov,
doi = {10.3389/fphar.2026.1985523},
url = {https://www.frontiersin.org/articles/10.3389/fphar.2026.1985523},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance
AU - Ru Lv
AU - Xiaoyu Xu
AU - Yulin Ji
AU - Jianxiang Wei
AU - Zuojian Zhou
AU - Kankan She
PY - 2026
DA - 2026/11/03
JO - Frontiers in Pharmacology
DO - 10.3389/fphar.2026.1985523
UR - https://www.frontiersin.org/articles/10.3389/fphar.2026.1985523
AB - IntroductionAdverse drug reaction (ADR) prediction remains challenging because spontaneous reporting data leave many exposure unit–ADR associations unobserved and provide limited support for candidate prioritization and mechanistic interpretation.MethodsWe developed MEG-HHGN, a mechanistic evidence-guided hyperbolic heterogeneous graph neural network. It represents exposure units as combinations of standardized drug names and routes of administration and integrates spontaneous reports with compounds, drug targets, ADR-related targets, and protein–protein interactions. Using 73,530 spontaneous reports collected in Jiangsu Province from 2010 to 2023, we evaluated MEG-HHGN against eight baseline methods through five-fold cross-validation.ResultsMEG-HHGN achieved AUC/AP/MCC values of 0.9550/0.9537/0.7917 in known-exposure-unit completion and 0.9090/0.9064/0.6587 in side-information-supported exposure-unit cold-start evaluation. Global scoring of unobserved exposure unit–ADR pairs produced a ranked candidate pool for pharmacovigilance review. Representative cases involving intravenous vinorelbine tartrate–myelosuppression and oral aspirin–proteinuria illustrated PPI-bridged and shared-target pathways, respectively.DiscussionMEG-HHGN integrates route-aware prediction, candidate prioritization, and pathway-based interpretation in a unified framework to support signal triage, mechanistic hypothesis generation, and evidence review in post-marketing pharmacovigilance.
SN - 1663-9812
ER -