EUVIMEDEUVIMED entdecken
Open Access

DrIVeNN: Drug Interaction Vectors Neural Network

Natalie Wang iD, Casey Overby Taylor

Journal of Computational Biology · 2025

DOI: 10.1089/cmb.2025.0079 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Computational Biology, 2025. Verfasst von Natalie Wang und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science, Johns Hopkins University Whiting School of Engineering, Baltimore, Maryland, USA. Hauptthemen laut MEDLINE: Neural Networks, Computer, Drug-Related Side Effects and Adverse Reactions. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 26 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NHGRI NIH HHS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Polypharmacy, the concurrent use of multiple drugs to treat a single condition, is common in patients managing multiple or complex conditions. However, as more drugs are added to the treatment plan, the risk of adverse drug events (ADEs) rises rapidly. Because it is impractical to test every possible drug combination during clinical trials, many serious polypharmacy ADEs (also known as drug-drug interactions or DDIs) only become known after the drugs are in use. This issue is prevalent among older adults with cardiovascular disease (CVD), where polypharmacy and ADEs are common. In this research, our primary objective was to identify key drug features and build and evaluate a model to predict DDIs. Our secondary objective was to assess our model on a domain-specific case study. We developed a two-layer neural network that incorporated drug features such as molecular structure, drug-protein interactions, and mono-drug side effects (drug interaction vectors neural network [DrIVeNN]) using publicly available side effect databases. It performed moderately better than state-of-the-art models such as DGNN-DDI, KGDDI, and NNPS. DrIVeNN had average area under the Receiver Operating Characteristic curve (AUROC) and area under the precision-recall curve (AUPRC) scores of 0.934 and 0.920, respectively, compared to the best-performing baseline model, DGNN-DDI, which had scores of 0.919 and 0.904. We also conducted a domain-specific case study centered on CVD treatment, and there was a significant increase in performance from the general model. We observed an average AUROC for CVD DDI prediction of 0.979. This research contributes to the advancement of predictive modeling techniques for polypharmacy ADEs and indicates the strong potential of domain-specific models.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Neural Networks, ComputerDrug-Related Side Effects and Adverse ReactionsHumansDrug InteractionsPolypharmacyComputational BiologyCardiovascular Diseases

Stichwörter der Autor:innen

adverse drug eventsdrug–drug interactionsneural networkspolypharmacy

Autor:innen und Institutionen

  • Department of Computer Science, Johns Hopkins University Whiting School of Engineering, Baltimore, Maryland, USA.
  • Institute for Computational Medicine, Johns Hopkins University, Baltimore, Maryland, USA.
  • Departments of Medicine and Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA.

ORCID-Profile: Natalie Wang

Förderung

  • NHGRI NIH HHS

Zitieren

APA

Wang, N., & Taylor, C. O. (2025). DrIVeNN: Drug Interaction Vectors Neural Network. Journal of Computational Biology. https://doi.org/10.1089/cmb.2025.0079

Vancouver

Wang N, Taylor CO. DrIVeNN: Drug Interaction Vectors Neural Network. Journal of Computational Biology. 2025. doi:10.1089/cmb.2025.0079

BibTeX
@article{Wang2025,
  author = {Wang, Natalie and Taylor, Casey Overby},
  title = {DrIVeNN: Drug Interaction Vectors Neural Network},
  journal = {Journal of Computational Biology},
  year = {2025},
  doi = {10.1089/cmb.2025.0079}
}
Kostenlos starten →