We present a calibrated model to scale-up a roller compaction process using minimal experimental data at small scales. The compression properties of three API-laden blends are evaluated for various drug loadings using a laboratory-scale roller compaction simulator (Int J Pharm 269:403–15, 2004). Using the generated blend compression properties, a calibrated model combining the Reynolds roller compaction model (Comput Chem Eng 34:1049–57 2010) and ribbon solid fraction measurements is developed to predict the ribbon solid fraction as a function of the roller compaction conditions (roll force and roll gap). The predictions of this model are compared with measurements from ribbons generated on the Gerteis minipactor at various scales using appropriate methods. The predicted solid fractions achieve good agreement with measured solid fractions for the three drug products. Further, the calibrated model is extended by developing a parametrized regression model where several API properties, formulation properties, and ribbon solid fraction predictions from the Reynolds model are taken as features to generate ribbon solid fractions for partially characterized active blends. The predictions of the hybrid model agree reasonably with measurements on ribbons produced on a Gerteis 3-W-Polygran. This hybrid model can then be deployed to run digital Design of Experiments (DOEs), which is then used to define operating ranges for the roller compaction of two drug products. This method enables the application of digital tools to aid the scale-up of fast-moving drug product processes using minimal experimental data.