AURIV Conference Abstracts

Abstract 1: ISMB/ECCB 2026

Title: AURIV: A Knowledge Graph-Based Platform for Systematic Drug Repurposing Discovery Using Relational Graph Neural Networks

Authors: AURIV Healthcare Intelligence Research Team

Track: Disease Models & Epidemiology / Systems Biology

Abstract (250 words max)

Motivation: Drug repurposing—identifying new therapeutic applications for approved drugs—offers a promising strategy to reduce drug development costs and timelines. However, the combinatorial space of drug-disease pairs makes systematic exploration challenging. Knowledge graphs provide a structured framework for integrating heterogeneous biomedical data and enabling computational drug repurposing discovery.

Results: We present AURIV, an AI-driven platform that integrates data from eight authoritative sources (DrugBank, DisGeNET, STRING, OMIM, ChEMBL, ClinicalTrials.gov, UniProt, and PubMed) into a comprehensive knowledge graph containing 6,777 nodes (427 drugs, 601 diseases, 5,749 proteins) and 47,832 weighted edges. Using Relational Graph Convolutional Networks (R-GCN) with DistMult scoring, AURIV predicts novel drug-disease associations with mechanistic interpretability through protein target pathway analysis. Cross-validation achieves AUROC of 0.891 and AUPRC of 0.847. Temporal validation shows 60.5% of drug-disease associations established between 2020-2025 were predicted within top-500 candidates. We demonstrate utility through case studies including metformin, where AURIV identified 12 repurposing opportunities supported by 47 active clinical trials. Predictions are stratified into evidence grades (A/B/C) based on clinical trial status, shared protein targets, and literature support. AURIV identified 2,847 novel drug-disease associations with grade B or higher lacking current clinical investigation, representing opportunities for therapeutic development.

Availability: Partnership inquiries: auriv@somasoft.com

Contact: auriv@somasoft.com

---

Abstract 2: AMIA Annual Symposium 2026

Title: AI-Enabled Drug Repurposing: A Knowledge Graph Approach for Accelerating Therapeutic Discovery

Authors: AURIV Healthcare Intelligence Research Team

Category: Artificial Intelligence / Clinical Informatics

Abstract (400 words)

Background: Traditional drug development requires 10-15 years and over $2.6 billion per approved therapy, with 90% of candidates failing during clinical development. Drug repurposing leverages existing approved drugs for new therapeutic indications, potentially reducing development time to 3-5 years and costs by up to 90%. We developed AURIV, an artificial intelligence platform to systematically identify drug repurposing opportunities through knowledge graph analysis.

Methods: AURIV integrates data from eight biomedical databases into a heterogeneous knowledge graph containing 6,777 nodes (427 FDA-approved drugs, 601 diseases, 5,749 protein targets) and 47,832 edges representing six relationship types (TARGETS, TREATS, ASSOCIATED_WITH, INTERACTS_WITH, CONTRAINDICATES, CAUSES). Node embeddings are learned using Relational Graph Convolutional Networks (R-GCN) with relation-specific transformations. Drug-disease link prediction employs DistMult scoring with negative sampling. Predictions are stratified into evidence grades: A (active Phase II/III trials), B (Phase I or strong mechanistic rationale), and C (computational prediction only). Validation included 5-fold cross-validation and temporal holdout testing.

Results: AURIV achieved AUROC of 0.891 (95% CI: 0.868-0.914) and AUPRC of 0.847 (95% CI: 0.816-0.878) in cross-validation. Temporal validation demonstrated that 60.5% of new drug-disease associations established between 2020-2025 appeared within top-500 predictions per drug. Case study analysis of metformin identified 12 repurposing opportunities across therapeutic areas including oncology, neurology, and metabolic diseases, with 47 supporting clinical trials. Platform-wide analysis identified 2,847 novel drug-disease associations with evidence grade B or higher that lack current clinical investigation, including several subsequently validated by clinical trials or regulatory approval.

Conclusion: AURIV demonstrates that knowledge graph-based approaches can systematically identify clinically relevant drug repurposing opportunities with mechanistic interpretability. The platform's evidence grading system enables prioritization of candidates for experimental validation. Particularly for rare diseases where 95% lack approved treatments, drug repurposing may represent the most viable path to therapeutic development. Future work includes integration of multi-omics data and real-world evidence from electronic health records.

Implications: AI-enabled drug repurposing can accelerate therapeutic discovery for conditions with unmet medical needs while reducing development costs and timelines.

---

Abstract 3: Drug Discovery & Development Summit 2026

Title: From Data to Drugs: An AI Platform for Systematic Drug Repurposing Across 600+ Diseases

Authors: AURIV Healthcare Intelligence

Format: Poster Presentation / Platform Demo

Abstract (300 words)

Drug repurposing represents a high-value strategy for pharmaceutical R&D, offering reduced risk, shorter timelines, and lower costs compared to de novo drug discovery. However, identifying promising repurposing candidates requires systematic analysis of drug mechanisms, disease biology, and clinical evidence across thousands of potential combinations.

AURIV is an AI-driven platform that accelerates drug repurposing discovery through comprehensive knowledge graph integration and graph neural network-based prediction. The platform integrates data from DrugBank, ChEMBL, DisGeNET, STRING, ClinicalTrials.gov, and PubMed into a unified knowledge graph containing:

- 427 FDA-approved drugs with established safety profiles - 601 human diseases across all major therapeutic areas - 5,749 protein targets with drug and disease associations - 47,832 weighted relationships capturing diverse evidence types

The AURIV AI employs Relational Graph Convolutional Networks to learn embeddings that capture complex drug-target-disease relationships. Link prediction identifies novel drug-disease associations with confidence scores and mechanistic interpretability through shared protein target analysis.

Key platform capabilities: - Evidence-graded predictions (A/B/C) based on clinical trial status and mechanistic plausibility - Validated predictive performance (AUROC 0.89, 60% temporal prediction accuracy) - Comprehensive target pathway analysis for mechanistic hypotheses - Clinical trial landscape mapping for competitive intelligence

Case study: Metformin analysis identified 12 repurposing indications with varying evidence levels, including cancer (23 active trials), PCOS, obesity, Alzheimer's disease, and aging. Shared AMPK-mTOR pathway modulation provides mechanistic rationale across diverse therapeutic applications.

AURIV offers partnership opportunities for pharmaceutical companies seeking to expand approved drug indications, biotech companies prioritizing development pipelines, and rare disease foundations identifying treatment candidates for underserved patient populations.

Contact: auriv@somasoft.com

---

Target Conferences 2026

| Conference | Deadline | Date | Location | Fit | |------------|----------|------|----------|-----| | ISMB/ECCB 2026 | ~Feb 2026 | July 2026 | TBD | High - Computational biology | | AMIA Annual | ~Mar 2026 | Nov 2026 | Washington DC | High - Clinical informatics | | Drug Discovery Summit | Rolling | Various | Various | High - Industry audience | | BIO International | ~Mar 2026 | June 2026 | Boston | Medium - Partnering focus | | ASHG Annual Meeting | ~June 2026 | Oct 2026 | TBD | Medium - Genetics focus | | ACS Spring Meeting | ~Oct 2025 | Mar 2026 | TBD | Medium - Chemistry focus | | DIA Annual Meeting | ~Jan 2026 | June 2026 | TBD | Medium - Regulatory focus |

---

Poster Design Notes

Title Panel: - AURIV logo - Title in large font - Author affiliations - Contact QR code

Main Panels: 1. Introduction & Problem Statement 2. Knowledge Graph Architecture (with visualization) 3. AI Methodology (R-GCN diagram) 4. Validation Results (AUROC curves) 5. Case Study: Metformin (table + network) 6. Novel Predictions (top 10 table) 7. Conclusions & Future Work 8. Partnership Opportunities

Design: - AURIV brand colors (blue gradient) - Clean, professional layout - Knowledge graph visualization as centerpiece - QR code linking to preprint