At the National University of Singapore, a team of researchers has done something that would have seemed impossible just a decade ago: they screened nearly 3,000 existing drugs against thousands of disease-related proteins in weeks, using artificial intelligence and molecular simulations to pinpoint a humble vitamin as a powerful ally in diabetic wound healing.
Diabetic foot ulcers are notoriously stubborn wounds. Unlike injuries on unaffected skin, they resist healing because multiple biological processes go haywire simultaneously—inflammation spirals, tissue repair stalls, cell growth falters. A drug that addresses one problem might worsen another. Finding an existing medication that could help felt less like solving a puzzle and more like searching for a needle in a haystack the size of a continent.
The multidisciplinary team led by Professor Giorgia Pastorin from the NUS Department of Pharmacy and Pharmaceutical Sciences, working with Associate Professor Chen-Hua Yeow from Biomedical Engineering and Associate Professor Min-Yen Kan from Computer Science, decided to build a better haystack-searcher. They developed an AI-guided workflow that combined three distinct layers of evidence: artificial intelligence scanning of scientific literature, computational chemistry modeling of molecular interactions, and traditional laboratory validation.
The scope was ambitious. The team mapped 2,989 existing drugs against 8,739 proteins linked to diabetic wound healing. AI algorithms combed through published research to identify which drugs might influence these proteins. Computational chemistry then calculated the strength of molecular interactions between the most promising candidates and their target proteins. These two layers of screening winnowed millions of possible combinations down to just 35 candidate drugs and 50 key proteins worthy of closer examination.
One name rose to the top of the ranked list: folic acid. It's a vitamin found in leafy greens and multivitamins, sold in drugstores for pennies, used as a dietary supplement by millions—but never considered a treatment for diabetic foot ulcers. When the NUS team tested folic acid in laboratory experiments using human skin cells, the results vindicated their computational predictions. Folic acid significantly improved wound closure in treated cells compared with untreated controls, suggesting a genuine therapeutic mechanism hiding in plain sight.
What makes this work genuinely remarkable is the efficiency gain. The integrated approach shrank the timeline from literature review to laboratory testing by more than 70 percent compared with conventional drug-discovery methods. Dr. Zhang Ziyang, the publication's first author, explains the architecture that makes this work: "AI helps identify possible biological directions, computational chemistry examines the molecular interactions, and the combined scoring aligns these two layers into testable priorities. Laboratory validation then closes the loop."
The research, published in ACS Nano Medicine, represents more than a single promising finding about folic acid. It demonstrates how AI and molecular simulation can uncover therapeutic connections that traditional research might overlook—hidden links between diseases and treatments that exist in plain sight but require machine learning to connect across the scientific literature at scale. For diabetic patients struggling with foot ulcers that refuse to heal, the possibility of a safe, inexpensive, widely available vitamin offering real therapeutic benefit is particularly welcome news. The question now is whether clinical trials will confirm what the laboratory has suggested.