At Chalmers University of Technology in Gothenburg, Sweden, researchers have developed an AI model that simulates molecular behavior at speeds that would have seemed impossible just years ago—10,000 times faster than the conventional methods that have been the scientific standard for decades. The breakthrough, published in Science Advances, promises to reshape how scientists discover and test new drugs, potentially shaving years off a process that currently consumes more than a decade from initial concept to patient care.
The challenge that molecular scientists have long faced is fundamentally about time and scale. When researchers use traditional molecular dynamics simulations, they must calculate the forces between every atom, then move them forward in tiny increments—each step lasting just one femtosecond, a millionth of a billionth of a second. Because the biological processes that matter for drug development unfold over much longer timescales, billions of these minuscule steps become necessary, making the simulations so computationally demanding that they can take weeks or months to complete for a single molecule.
The research team, led by Simon Olsson, an associate professor in the Department of Computer Science and Engineering, took a fundamentally different approach. Rather than simulating every step in sequence, their AI model learns the underlying rules that govern how molecules move and change over time. "What sets our AI model apart is that it learns the underlying dynamics over longer timescales," Olsson explains. "It not only provides insights into the shapes that molecules take on, but also into how quickly and through which pathways these molecular transitions occur."
To develop and validate their approach, the researchers trained the model on more than 12,500 organic molecules—compounds containing carbon, nitrogen, hydrogen, and oxygen—along with over 1,000 short peptides, which are chains of amino acids that form proteins. The model learned the fundamental patterns of molecular behavior from these examples, then proved capable of predicting how entirely new molecules it had never encountered would behave. When the team compared their AI predictions to results from standard numerical algorithms, the outputs aligned perfectly, confirming that the model's speed hadn't come at the cost of accuracy.
The implications for drug development are substantial. Currently, the early stages of drug discovery involve screening thousands of molecular candidates, with only a fraction advancing to the next phase of testing. Scientists must pour significant resources into expensive laboratory work and extended computational simulations just to narrow down which candidates are worth pursuing further. By accelerating these simulations by 10,000-fold, the AI model could allow researchers to rapidly identify the most promising drug candidates before investing in costly wet-lab experiments.
Olsson describes the innovation as enabling researchers to "jump between scenes in molecular movies, instead of watching every frame in sequence." The practical applications extend beyond speed; in the laboratory, scientists will eventually be able to measure specific molecular properties—whether they dissolve well in solution, or whether they can pass through cell membranes—informed by AI predictions that took seconds rather than months to generate.
The team emphasizes that one of the model's greatest strengths is its generalizability. Because it learned fundamental rules rather than memorizing specific molecular systems, it can reliably predict behavior for molecules far beyond its training set. As drug development continues to grapple with timelines and costs, this Swedish innovation offers a tangible path toward faster, more efficient discovery of the medicines that patients need.