https://www.jmir.org/2026/1/e101366
QT:{{”
Mark Gerstein, PhD—professor of biomedical informatics, molecular biophysics, and computer science at Yale University—said that this paradox helps explain why enthusiasm around AI should be tempered with realism. “There is a lot of excitement with AI in drug discovery,” Gerstein said, “but the reality is that we have not been as successful at discovering drugs as we would have hoped.”
…Gerstein also points to equivariant neural networks as a promising advance in AI-driven drug discovery. Since molecules obey physical laws and can rotate, stretch, or change their conformation, predictive models must account for these transformations. Equivariant
architectures are designed to preserve these relationships, making them especially valuable for molecular modeling and binding prediction “}}
Cuffari, B. (2026). From virtual molecules to clinical trials: How AI is reshaping preclinical drug discovery. Journal of Medical Internet Research, 28, e101366. https://doi.org/10.2196/101366
linkstream2 microblog 