In a recent review from Xidian University in China, researchers have shown how computer-based methods can help find new uses for existing medicines, offering hope for faster and cheaper treatments. The study highlights a clear health policy challenge: making safe, effective therapies available quickly without the huge cost and delay of traditional drug development.

“These advanced tools allow us to see drug–disease relationships that were previously hidden in plain sight,”  Prof. Liang Yu, the lead author, explains

Developing a new medicine from scratch typically takes 13 to 15 years and can cost around $2–3 billion, with about a 90% failure rate in clinical trials. According to data from the Tufts Center for the Study of Drug Development (2016), this high cost and slow pace leave many diseases without effective treatments, especially rare or neglected conditions.

By contrast, drug repurposing uses medicines already approved for one illness to treat another. Because these drugs have known safety records, clinical trials can be shorter, cheaper, and less risky. A 2021 Nature Reviews Drug Discovery paper notes that repurposed drugs can often reach approval in 3–6 years instead of 15.

The Xidian University review focuses on “in silico” (computer-based) tools. These include: neural networks, where artificial intelligence systems that learn patterns from data to predict new drug–disease matches. In the Xidian benchmark study covering 663 drugs and 409 diseases, neural networks were the most accurate.

Moreover, matrix-based methods provide faster models that use mathematical structures to find relationships. Slightly less accurate but very efficient added with recommendation algorithms, or systems similar to those used in online shopping, matching drugs and diseases by their similarities and text mining, whereor tools that scan large numbers of scientific papers to extract hidden connections.

These techniques work best when they use high-quality, structured data. For this, the researchers built a comprehensive database from trusted sources like DrugBank, OMIM (Online Mendelian Inheritance in Man), and PubChem. They even added carefully designed “negative samples” to avoid false results—a key concern in artificial intelligence research.

Prof. Yu emphasizes: “By mapping strengths and limitations across diverse computational frameworks, we’re laying out a clear roadmap for how digital repurposing can improve medicine.”

Experts say these methods could help health systems respond faster to new diseases, as seen during COVID-19, when repurposing efforts rapidly tested existing antivirals. The World Health Organization’s COVID-19 Solidarity Trial used similar ideas, showing that global cooperation and data-sharing can accelerate progress.

For policymakers, the takeaway is clear: supporting data sharing, building better databases, and funding collaborative research can cut costs and speed up treatment availability. While these methods aren’t a silver bullet, they could reduce health inequalities by making effective therapies accessible to more people worldwide.