Cellarity published a manuscript in Science describing a framework that fuses transcriptomic datasets with AI modeling to discover ‘cell state‑correcting’ medicines. The company outlined methods to read disease‑relevant cell states and predict compounds that revert pathological states toward healthy programs. The work details a computational and experimental pipeline that integrates multi‑omics and machine learning to suggest therapeutic interventions that change cellular phenotypes rather than only targeting single genes. Cellarity said the approach shortens hypothesis generation and prioritizes targets with system‑level effects, citing internal validations and case examples. For drug developers, the Science paper signals a growing shift toward cell‑state phenotypic discovery enabled by large datasets and foundation models. Academic and industry groups will evaluate reproducibility, translational robustness and how the approach complements conventional target‑based and high‑throughput screens.