Mechanistic modeling for cell culture process of trastuzumab and intelligent algorithm comparison

Mechanistic modeling, which describes cellular metabolic behavior through mathematical equations, is an important theoretical tool for optimizing and predicting biological processes. In this study, a cell culture mechanistic model for biosimilar trastuzumab was constructed based on the Monod growth metabolism model. To address the high complexity of differential equation parameter fitting and the tendency to fall into local optima, particle swarm optimization (PSO), grey wolf optimization (GWO), and differential evolution (DE) algorithms were introduced for model parameter fitting. The results show that GWO outperforms the others in global search capability and fitting accuracy, demonstrating higher stability and adaptability in handling multi-parameter fitting problems. Furthermore, improvements were made to the position and velocity update methods of the GWO algorithm. The improved grey wolf optimization (IGWO) significantly enhanced the model's predictive ability, enabling accurate predictions of cell density, metabolite concentration, and antibody titer. This study provides important theoretical support for improving the production efficiency of biopharmaceuticals, reducing development costs, and optimizing cell culture processes.