Scientists are using AI to speed up drug repurposing, expanding the treatment possibilities for people with rare diseases and few options
Joseph Coates, left, owes his life to an A.I. model developed by Dr. David Fajgenbaum and the rest of his team.Hannah Yoon for The New York Times Bluesky: @rjcc , @karissabe , @nochoicebutaction , and @novaforhealth . Threads: @owen.sakawa X: @reidhoffman , @kempnerinst , @erictopol , @celinegounder , @clintonglobal , @katie_haun , @jondavidson_ , @davidfajgenbaum , @tedchris , and @korthlab_hhu . LinkedIn: Luca Foschini Bluesky: Richard Lawler / @rjcc : It's interesting that this doesn't seem to discuss the process at work in deeper detail than “we ran a search query” and “using ai” Karissa Bell / @karissabe : This is actually a pretty interesting use of AI. Also notable that “drug repurposing is not likely to prove lucrative for any one party” — www.nytimes.com/2025/03/20/w... [image] @nochoicebutaction : Many drugs do more than one thing, Dr. Might said. Their additional features sometimes get characterized as side effects. “If you comb through enough drugs, you eventually find the side effect you're looking for,” he said, “and then that becomes the main effect.” — www.nytimes.com/2025/03/20/w... @novaforhealth : Need some good news? Read the latest piece from Nova Media Fellow Kate Morgan in @nytimes.com about how doctors are using AI to find new uses for old medicines—and save lives. — www.nytimes.com/2025/03/20/w... #ai #ArtificialIntelligence #AIinHealth #DigitalHealth Threads: Owen Sakawa / @owen.sakawa : AI just saved a life when doctors told him he was going to die. But this miracle could kill the next patient. → Scanned 100M+ records in seconds. → Found a forgotten drug that worked. → BUT... 1 hallucination = 1 funeral. Should AI be allowed in hospitals without guardrails? … X: Reid Hoffman / @reidhoffman : It is inspiring to see AI already making an impact in medicine and saving lives. Good read, if you're looking for some positive news. https://www.nytimes.com/... @kempnerinst : #KempnerInstitute & @harvardmed faculty @marinkazitnik speaks to the @nytimes about her work using #AI to repurpose drugs for new treatments: https://nytimes.com/.... Read our coverage of Zitnik's groundbreaking work: https://kempnerinstitute.harvard.edu/ ... @DrYohanJohn Eric Topol / @erictopol : “In labs around the world, scientists are using A.I. to search among existing medicines for treatments that work for rare diseases.” And saving some lives. @DavidFajgenbaum https://www.nytimes.com/... [image] @celinegounder : .@Penn @EveryCure's Dr. @DavidFajgenbaum, who we featured in our story about drug repurposing & the use of leucovorin for treatment of autism, was in the @nytimes today for his work on repurposing a drug for a rare blood disorder. https://www.nytimes.com/... @clintonglobal : .@NYTimes: Doctors Told Him He Was Going to Die. Then A.I. Saved His Life. At #CGI2022, Dr. David Fajgenbaum's nonprofit, Every Cure, committed to create an open-source database of drug treatments for rare diseases. This week, the New York Times reported on the success of his Kathryn Haun / @katie_haun : Through force of will @DavidFajgenbaum found the cure for his own life-threatening disease. But knowing David it's no surprise that was just the beginning, now he's using AI to do that for other rare diseases too. https://www.nytimes.com/... Jon Davidson / @jondavidson_ : Through his nonprofit, @EveryCure, Dr. @DavidFajgenbaum is turning heartbreak into hope. By repurposing existing drugs to treat rare diseases, he's quietly revolutionizing medicine—and now he's harnessing A.I. to move even faster. Worth a read: https://www.nytimes.com/... David Fajgenbaum / @davidfajgenbaum : We are thrilled to share that the @nytimes has just featured @EveryCure and our work using repurposed drugs to save lives. Click the link below to read the full article! https://www.nytimes.com/... Chris Anderson / @tedchris : This is a truly incredible story. As inspiring as they come. Also, David will be speaking at TED next month.... https://www.nytimes.com/... @korthlab_hhu : “If you comb through enough drugs, you eventually find the side effect you're looking for,” he said, “and then that becomes the main effect.” https://www.nytimes.com/... LinkedIn: Luca Foschini : After yesterday's devastating news of the $25M cut to the Department of Defense Neurofibromatosis research program — a bipartisan initiative …
Context & Ripple Effects
AI drug-development coverage has largely focused on finding new compounds and on emergency-era efforts such as BenevolentAI’s search for new uses of existing medicines during COVID-19. This case shifts the emphasis to systematically matching already-known drugs with underserved conditions, where the economics of a single treatment can be weak.
EveryCure’s planned open-source database makes the project more than a one-off clinical lead: it proposes shared discovery infrastructure for rare-disease treatment matching. That sits alongside a broader push to apply AI across drug R&D, including AI-driven discovery and development at Terray Therapeutics.
First-order effects
- Patients with rare diseases may gain a faster route to candidate treatments from the existing drug catalog, while clinicians and researchers receive a larger set of leads to evaluate.
- EveryCure and Fajgenbaum’s team gain a concrete demonstration of their model’s clinical relevance and a basis for building an open database of potential drug-disease matches.
Second-order effects
- Because repurposing is unlikely to be highly lucrative for a single party, open data and nonprofit-style coordination could become more important than conventional drug-development incentives for surfacing these opportunities.
- The faster AI generates plausible matches, the greater the need for clinical review and guardrails; earlier coverage warned that medical AI can amplify overtesting and overtreatment when outputs are used without adequate oversight.
Third-order effects
- If shared treatment-matching databases prove useful, rare-disease research could move toward an AI-enabled commons in which the bottleneck shifts from finding candidates to validating, funding, and deploying them.
- The durable test for this model is whether AI-assisted leads can be integrated into accountable clinical workflows rather than treated as recommendations on their own.
The trend: AI in medicine is expanding from proprietary molecule discovery toward shared systems that search existing evidence and therapies for neglected clinical needs.