Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance
Frontiers in Pharmacology · 03.11.2026 · 17
DOI: 10.3389/fphar.2026.1985523 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Pharmacology, online erschienen am 03.11.2026. Verfasst von Ru Lv und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Artificial Intelligence and Information Technology, Nanjing University of Chinese Medicine, Nanjing, China. Die Arbeit nennt 63 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, National Social Science Fund of China, Social Science Foundation of Jiangsu Province.
Abstract (Originalsprache)
Introduction
Adverse 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.
Methods
We 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.
Results
MEG-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.
Discussion
MEG-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.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Artificial Intelligence and Information Technology, Nanjing University of Chinese Medicine, Nanjing, China.
- Jiangsu Engineering Research Center for Intelligent Health Services of Traditional Chinese Medicine, Nanjing University of Chinese Medicine, Nanjing, China.
- School of Management, Nanjing University of Posts and Telecommunications, Nanjing, China.
Förderung
- National Natural Science Foundation of China
- National Social Science Fund of China
- Social Science Foundation of Jiangsu Province
- Nanjing University of Chinese Medicine
- Jiangsu Provincial Department of Education
Zitieren
APA
Lv, R., Xu, X., Ji, Y., Wei, J., Zhou, Z., & She, K. (2026). Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance. Frontiers in Pharmacology, 17. https://doi.org/10.3389/fphar.2026.1985523
Vancouver
Lv R, Xu X, Ji Y, Wei J, Zhou Z, She K. Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance. Frontiers in Pharmacology. 2026;17. doi:10.3389/fphar.2026.1985523
BibTeX
@article{Lv2026,
author = {Lv, Ru and Xu, Xiaoyu and Ji, Yulin and Wei, Jianxiang and Zhou, Zuojian and She, Kankan},
title = {Hyperbolic heterogeneous graph learning for adverse drug reaction prediction, candidate prioritization, and pathway-based interpretation in pharmacovigilance},
journal = {Frontiers in Pharmacology},
year = {2026},
volume = {17},
doi = {10.3389/fphar.2026.1985523}
}
EUVIMED