imAgeScore, a Cell Painting-Based Predictor of Cellular Age for High-throughput Drug Screening Applications
Quantifying cellular age in vitro in a scalable and biologically meaningful manner is essential for the discovery of pharmacological interventions that modulate aging. We developed imAgeScore, a machine-learning model trained on high-content Cell Painting features to predict the phenotypic age of primary human dermal fibroblasts. imAgeScore correlates with chronological and DNA methylation-based age estimates and captures coordinated morphological changes across nuclear and cytoplasmic compartments. The model detected age acceleration during serial propagation and age reduction following partial reprogramming. Pharmacological interventions targeting distinct aging hallmarks induced predictable shifts in predicted age and enabled classification of damaging versus rejuvenating cellular states. Application of imAgeScore in an automated high-throughput screening pipeline identified candidate age-modulating compounds, revealed inter-individual variability in response magnitude, and detected additive rejuvenation effects in selected combinatorial treatments. Functional validation in a scratch wound assay confirmed enhanced cellular repair by leading candidates, supporting the biological relevance of morphology-derived age reduction. Together, these results demonstrate that image-based morphological profiling provides a scalable platform for quantifying cellular aging and screening for candidate rejuvenation interventions.