AI's latest trick: Repurposing old drugs for rare diseases
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Aïda Amer/Axios
A new artificial intelligence tool could supercharge efforts to find new uses for old drugs, particularly rare diseases without a Food and Drug Administration-approved treatment.
The big picture: The Harvard Medical School researchers behind the tool, called TxGNN, write today in Nature Medicine that it can identify candidates for 17,000 conditions — the largest number of diseases that any single AI model can handle to date.
- The team is making the tool available for free to spur more research on untreated or undertreated conditions that affect 300 million people worldwide.
Details: The tool has two features: one that identifies drug candidates and possible side effects and another that explains the rationale for the decision.
- In initial experiments, it identified drug candidates from nearly 8,000 medicines, including those that are already approved and those still in trials.
Between the lines: There are other tools on the market that ID drugs that can be repurposed. But the process often is "serendipitous and opportunistic," the researchers write.
- Researchers said they found this tool was 50% better on average at identifying drug candidates compared against the leading AI models and 35% more accurate in predicting what drugs would have potentially harmful effects.
Yes, but: The researchers acknowledge the tool's success is only as good as the medical knowledge it uses to derive conclusions.
- "Challenges such as data biases and potentially outdated information in the knowledge graph must be addressed," they wrote.
The bottom line: The AI model could provide a more cost-effective way to develop therapies than designing new drugs from scratch, said lead researcher Marinka Zitnik in a statement.
- "We foresee this model could help close, or at least narrow, a gap that creates serious health disparities," said Zitnik, an assistant professor of biomedical informatics in the Blavatnik Institute at Harvard Medical School.
