Temporal AI model predicts drivers of cell state trajectories across human aging
Year: 2026
Type: preprint
Abstract: Foundational AI models have recently shown promise for predicting the impact of perturbations on cell states. However, current models typically consider only one cell state at a time, limiting their ability to learn how cellular responses unfold over time, particularly across long trajectories such as diseases of aging. Here, we develop a temporal AI model, MaxToki, trained on nearly 1 trillion gene tokens including cell state trajectories across the human lifespan to generate cell states across long timelapses of human aging. MaxToki generalized to unseen trajectories through in-context learning and predicted novel age-modulating targets that were experimentally verified to influence age-related gene programs and functional decline in vivo. MaxToki represents a promising strategy for temporal modeling to accelerate the discovery of interventions for programming therapeutic cellular trajectories.
Authors: Javier Gόmez Ortega, Rangarajan D. Nadadur, Akira Kunitomi, Steven Kothen-Hill, Julian U. G. Wagner, Sarp D. Kurtoglu, Bumjoon Kim, Madigan M. Reid, Thomas Lu, Kaho Washizu, Lukas Zanders, Han Chen, Yujie Zhang, Sarah Ancheta, Sara Lichtarge, William A. Johnson, Chris Thompson, Dereck M. Phan, Alexis J. Combes, Andrew C. Yang, Neha Tadimeti, Stefanie Dimmeler, Shinya Yamanaka, Michael Alexanian, Christina V. Theodoris
Institutions: Gladstone Institutes, University of California, San Francisco, University of California San Francisco Medical Center, Nvidia (United States), Goethe University Frankfurt, University Hospital Frankfurt, University of California, Berkeley, Kyoto University, Center for iPS Cell Research and Application, Kyoto University
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7147583936