Researchers from Nanyang Technological University (NTU) and the Massachusetts Institute of Technology (MIT) have developed an artificial intelligence model that could revolutionize RNA medicine development by dramatically reducing the time needed to design effective drug delivery systems. The breakthrough, published in Nature Nanotechnology, introduces COMET, an AI "experiment simulator" that predicts optimal formulations for lipid nanoparticles (LNPs) - the microscopic vehicles that transport mRNA into cells.
Transforming Drug Development Timelines
COMET addresses one of the most complex challenges in RNA medicine: designing the optimal "vehicle" to deliver therapeutic RNA molecules safely to their cellular destinations. Traditional approaches require testing thousands of physical samples, a process that can take months. By identifying optimal designs computationally, COMET could reduce development timelines from months to weeks.
"This breakthrough showcases how AI can help make the development of new medicine faster, cheaper, and better," said Assistant Professor Alvin Chan from NTU's College of Computing and Data Science, who began the project as a postdoctoral researcher at MIT. "By drastically reducing the real-world experiments needed, we can get effective medicines to patients faster, and at lower cost."
Superior Performance Against Clinical Benchmarks
The AI model demonstrated remarkable predictive capabilities in validation studies. Several COMET-designed LNPs outperformed clinically approved benchmarks in both laboratory and animal studies, signaling a major advancement in RNA medicine delivery technology. The model found nanoparticle designs that exceeded existing clinical standards, representing a significant leap forward in the field.
Advanced AI Architecture for Complex Predictions
COMET is built on a transformer-based neural network architecture, similar to advanced language models, enabling it to detect complex relationships between nanoparticle design and delivery performance. The model was trained on one of the largest and most diverse datasets of lipid nanoparticle formulations ever compiled, capturing variations in materials, ingredient ratios, and manufacturing methods.
The research combined AI with high-throughput screening, creating an exceptionally rich dataset that enables COMET to explore a vast design space virtually. Instead of making and testing thousands of physical samples, scientists can input potential LNP "recipes" into COMET, which predicts their efficacy for different applications.
Solving the LNP Design Challenge
Lipid nanoparticles consist of four key components: cholesterol, a helper lipid, an ionizable lipid, and a lipid attached to polyethylene glycol (PEG). Each component can have multiple variants, creating an enormous number of possible combinations. Testing each mixture individually is impractical due to the sheer volume of possibilities.
"This is where COMET becomes a general-purpose technology for RNA medicines — whether you're targeting different cells or improving stability, the same system can guide you to the best solution," Chan explained. The model's application-flexible design allows it to adapt formulations for different organs or cell types, making it a potential general-purpose platform for RNA-based medicines.
Broad Therapeutic Applications
While pandemic preparedness represents a clear application, COMET's potential extends far beyond infectious diseases. The predictive power could help scientists develop RNA therapies that target specific organs or cell types, opening new possibilities for treating cancers and other complex diseases. The technology could also enhance the efficiency of existing RNA vaccines and enable new treatments for metabolic disorders such as obesity and diabetes.
Future Development and Platform Expansion
Chan's team is now developing an AI-powered platform that designs the most effective lipid nanoparticle for any chosen therapeutic target, paving the way for RNA medicines that can reach and treat diseases in specific organs or cell types with greater precision and safety. The team is also developing high-throughput experimental technologies that can test hundreds of RNA formulations in a single animal, accelerating the feedback loop between AI design and real-world validation.
The U.S. Advanced Research Projects Agency for Health (ARPA-H) is funding a multiyear research program at MIT to develop ingestible devices for oral delivery of RNA treatments and vaccines, highlighting the broader potential of AI in advancing RNA-based therapies and maximizing protein production for therapeutic applications.
