New uses for existing drugs are going unnoticed

How AI helps surface drug-repurposing hypotheses

Repurposing tools look for connections between compounds and diseases that no single dataset makes obvious. By integrating drug-target data, disease biology, gene-expression signatures, the literature, and sometimes real-world clinical data — frequently within a knowledge graph — they can propose that a compound acting on a particular mechanism might be relevant to a disease driven by that same mechanism. Some approaches match a drug's molecular signature against a disease signature; others reason over networks of relationships. The output is a ranked set of hypotheses with supporting evidence, not a conclusion. Because an approved compound's safety and pharmacology may already be characterized, a well-evidenced repurposing hypothesis can be attractive — but it still needs biological and clinical validation. The AI widens the search; scientists judge plausibility and own the decision to pursue a lead.

What to evaluate before buying repurposing AI

Evidence transparency is central: a repurposing suggestion is only actionable if you can see why it was made and judge whether the reasoning holds. Favor platforms that link each hypothesis to its underlying associations and sources, and ask how they distinguish mechanistically grounded connections from coincidental statistical overlap, which can generate plausible-looking but spurious hypotheses. Also weigh data breadth and freshness, since repurposing depends on connecting diverse sources, and consider practical constraints the software cannot see — intellectual-property status, formulation, and dosing differences between the original and proposed use. Check that the workflow lets your scientists triage and annotate hypotheses rather than handing back an opaque list.

How teams typically get started

A sensible entry point is testing the platform against known repurposing cases — compounds already found to work in a second indication — to see whether it would have surfaced those connections and how they rank against noise. Running it on a disease area your team understands well also reveals whether its top hypotheses are biologically credible before any is taken forward for validation.

AI Use Cases That Address This Problem

• Hit Identification & Lead Optimization
• Target Identification & Validation

Frequently asked questions

What is drug repurposing and why is it attractive?

Drug repurposing means finding new therapeutic uses for existing or previously studied compounds. It can be appealing because the safety profile and pharmacology are often already characterized, which may reduce some early risk — though a new indication still requires its own validation and clinical evidence.

How does AI find repurposing opportunities?

By integrating drug, target, disease, and molecular data — often as a knowledge graph — and looking for connections that link a known compound to a new disease through shared mechanisms or matching molecular signatures. The system generates and ranks hypotheses with supporting evidence for scientists to evaluate.

Are AI-generated repurposing hypotheses ready to act on?

No. They are starting points that require biological and clinical validation, and they can reflect coincidental data overlap rather than real mechanism. Practical factors the software may not capture — intellectual property, formulation, and dosing — also shape whether a hypothesis is worth pursuing, so treat the ranked list as decision support for expert review.

What should we ask a repurposing-platform vendor?

Ask what data sources are integrated and how current they are, how each hypothesis links to traceable evidence, how the platform separates mechanistic connections from statistical coincidence, and whether it would have surfaced known repurposing cases in a retrospective test. Confirm your scientists can inspect and triage the reasoning.

AI Vendors for This Problem

• Aqemia
• BenevolentAI
• Cellarity
• Chai Discovery
• Chemify
• Cradle Bio
• Deep Genomics
• Envisagenics
• Genesis Molecular AI
• Healx
• Iktos
• Immunai
• Insilico Medicine
• Isomorphic Labs
• NVIDIA BioNeMo
• Numerion Labs
• Owkin
• Recursion Pharmaceuticals
• Schrodinger
• Standigm
• Turbine
• Valo Health
• Variational AI
• XtalPi

Evidence & Outcomes