AI Joins the Quest to Find New Treatments for Rare Diseases

• The EU-funded DREAMS project uses artificial intelligence to accelerate drug discovery for five rare neuromuscular disorders.
• AI helps analyze massive datasets to identify existing drugs suitable for repurposing, significantly reducing the time and cost associated with drug development.
• Researchers are utilizing induced pluripotent stem cells (iPSCs) to model diseases in the lab and identify shared biological pathways across different conditions.
• The project is testing 2,700 existing drugs to address symptoms, with the ultimate goal of implementing "basket trials" to allow faster clinical testing for small patient populations.

AI in Drug Discovery

• Belgian AI company Kantify, led by CEO Ségolène Martin, pivoted its operations to human health after a personal health crisis in 2017.
• The company's platform, Sapian, processes data on molecules, proteins, and existing drug characteristics to generate highly accurate research hypotheses.
• This AI-driven approach targets the fact that only 5–6% of the world's 7,000–10,000 known rare diseases currently have approved therapies.

Research Methodology

• DREAMS focuses on disorders such as Duchenne muscular dystrophy and Emery–Dreifuss muscular dystrophy.
• By converting patient cells into iPSCs and subsequently into skeletal muscle tissue, the team creates a controlled environment to study disease mechanisms at the cellular level.
• Researchers, including project coordinator Xavier Nissan from the French institute I-Stem, are working to move beyond single-disease approaches by identifying common therapeutic targets.

Clinical Challenges and Future Outlook

• Clinical trials for rare diseases are notoriously difficult due to limited patient numbers; the team advocates for "basket trials" where one drug is tested across different diseases sharing a common biological cause.
• While the DREAMS project is funded through 2028, researchers remain cautious about timelines for human clinical results and emphasize that AI results still require real-world testing.