Alyssa Burke
Robust process development enhances product quality while reducing time and cost across the product lifecycle. Systematic approaches are necessary to fully understand processes and control sources of variability; however, challenges increase when entering the field of autologous or personalized cell and gene therapies (ex. CAR-T) where each patient brings inherent biological variability. This presentation utilizes a simulated case study that combines Design of Experiments (DOE), mechanistic modeling, machine learning models, and Monte Carlo simulation to illustrate how complex processes can be analyzed in silico alongside benchtop experiments.A screening DOE study defined an efficient experimental space while mechanistic modeling generated critical quality attribute (CQA) outcomes to complement benchtop experimentation. These results trained regression and random forest models which were fed into Monte Carlo simulations. The simulations then quantify how patient-specific versus process-controlled variability contributes to overall outcome variance. Simulations can be repeated under different parameter constraints to ensure variability in patient-inputs can still lead to CQAs that are within acceptable limits.This integrated approach provides a more robust process development than DOE alone. This framework can aid in guiding process optimization and risk assessments both in early-stage process development and in continuous improvement during commercial production.