HyperDC: A Non-Uniform Hypergraph Framework for Dual- and Higher-Order Drug Combination Recommendation Across Diverse Complex Diseases

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Up to 12.8 percentage-point improvement (unified benchmark); up to 30.0 percentage-point improvement (disease-specific tasks); 66.7% in vitro hit rate (4/6 tested MASH combinations showed synergistic lipid-lowering and anti-inflammatory effects); 70% of top 10 single-drug MASH candidates supported by recent literature

• Primary Outcome: HyperDC outperformed representative computational comparator methods by up to 12.8 percentage points in unified dual-drug benchmark tasks and up to 30.0 percentage points in disease-specific prediction tasks; superior recall performance across multiple top-ranked recall settings in fixed-anchor clinical ranking tasks with preferential recall of FDA-approved dual-drug and clinically supported three-drug combinations

• Nnt Or Sensitivity: Not applicable in conventional clinical sense; recall metrics used as primary performance indicators in computational benchmarking; specific recall values at top-k thresholds not provided in abstract

• Confidence Interval: Not reported

Clinical Application

As a computational tool, HyperDC would require access to curated drug-disease knowledge graphs and significant computational infrastructure. Deployment in a clinical or pharmaceutical research setting would require further validation, user interface development, and integration with existing drug development workflows. Direct clinical use by prescribers is not feasible or appropriate at this stage of development. MASH (formerly NAFLD/NASH) is a significant and growing burden in Australia, with an estimated prevalence of 25–30% of the adult population having NAFLD and a substantial subset progressing to MASH. No TGA-approved pharmacotherapy specifically for MASH existed at the time of publication, making combination therapy prioritisation tools potentially relevant to Australian drug development pipelines. The PBS does not currently list MASH-specific combination regimens. RACGP guidelines emphasise lifestyle modification as first-line management; pharmacological combination strategies remain investigational. HyperDC's FDA-approved combination validation benchmark has partial relevance to TGA-approved drugs, though regulatory approval pathways differ. Australian researchers and pharmaceutical companies could potentially utilise such a framework in preclinical drug discovery, but clinical translation would require HREC-approved trials and TGA regulatory engagement. The tool does not address Australian-specific disease prevalence patterns, Indigenous health considerations, or PBS cost-effectiveness requirements. Patients with complex multi-target diseases requiring combination pharmacotherapy, including but not limited to MASH, oncology, and other conditions with established polypharmacy rationale; currently applicable only as a drug discovery/prioritisation research tool, not for direct clinical decision-making

Abstract

Complex diseases are commonly characterized by dysregulation across multiple targets and pathways, making drug combinations an important therapeutic strategy. This creates a growing need for computational methods to prioritize candidate combinations, yet current methods face several practical challenges, including cross-disease application, unified prediction of dual- and multidrug combinations, lack of reliable negative labels, and prediction for low-resource diseases. In this study, we proposed HyperDC, a unified cross-disease drug combination recommendation framework. HyperDC constructs a nonuniform hypergraph based on clinical and knowledge-driven drug-disease association data, representing single drugs and dual- and multidrug combinations in a unified modeling space. It further integrates knowledge graph pretraining and adversarial negative sampling to enhance model discrimination across tasks with different difficulty levels. In the unified dual-drug benchmark and disease-specific prediction tasks, HyperDC outperformed representative methods by up to 12.8 and 30.0 percentage points, respectively. In fixed-anchor clinical ranking tasks, HyperDC showed superior performance across multiple top-ranked recall settings and preferentially recalled FDA-approved dual-drug combinations and clinically supported three-drug combinations. In the data-sparse metabolic dysfunction-associated steatohepatitis (MASH) scenario, HyperDC completed a full workflow from anchor-drug identification and combination partner prioritization to in vitro experimental validation: 70% of the top 10 single-drug candidates were supported by recent literature, and 4 of 6 tested combinations showed synergistic lipid-lowering and anti-inflammatory effects, yielding a hit rate of 66.7%. Overall, HyperDC may narrow the screening space, improve prioritization efficiency, and provide methodological support for the development of combination therapy strategies for complex diseases.

References

1. 1.Yu, H., Chen, X., Song, W., Xiong, L., Li, X., Wang, Z., Pan, F., Li, W., Liu, G., Gao, F., & Tang, Y. (2026). HyperDC: A non-uniform hypergraph framework for dual- and higher-order drug combination recommendation across diverse complex diseases. Journal of Chemical Information and Modeling. Advance online publication. https://doi.org/10.1021/acs.jcim.6c01470