Answers are grounded in this paper and cite their sources.
OBJECTIVE: Ensuring medication safety for patients with multi-morbidity, who are often prescribed multiple drugs concomitantly (i.e., polypharmacy), is challenging due to the high risk of drug-drug interactions (DDIs) and accumulation of side-effects. Leveraging large datasets in the form of Knowledge Graphs (KGs) can improve knowledge on drug safety by modeling complex relationships among biomedical entities such as drugs, diseases, proteins and side effects, allowing for the prediction of novel relations. However, it remains unclear to what extent existing KGs adequately capture the polypharmacy context. Therefore, this systematic review identifies KGs constructed and/or utilized for predicting side effects and DDIs in the polypharmacy context, and evaluates their reporting quality. METHODS: Publications were identified through a structured search through DBLP, MEDLINE, and the ACM Guide to Computing Literature up to January 2026, and supplemented by snowballing. We included peer-reviewed papers presenting KGs developed for and/or used in the context of polypharmacy-related predictions and containing polypharmacy knowledge. We excluded KGs missing both data and code. We adapted and extended a KG evaluation framework to objectively assess the reporting quality of KGs through a review of their documentation from both a technical and a medical domain-specific perspective. RESULTS: Ten studies were included following full-text screening. We provide an overview of relevant statistics, contents, and limitations using our KG evaluation framework. Our study revealed inconsistent technical documentation on KG construction and KG metadata, as well as a lack of knowledge on granular pharmacogenetics and (patho)physiological pathways within the included KGs. CONCLUSION: Our results show a need for improved quality of reporting, and the incorporation of fine-grained knowledge on pharmacogenetics and (patho)physiological pathways to ensure the long-term re-usability and relevance of KGs in the context of polypharmacy-related predictions. We also recommend to involve medical experts in KG development in this domain.
No takes yet. Share an insight, caveat, or question.
VOS et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: