Mechanistic Digital Model for Real-Time Monitoring and Optimization of Adenovirus Production

Adenoviral vectors are a critical vaccine platform, with billions of vaccine doses administered during the COVID-19 pandemic. In addition, they are promising therapeutics for cancer treatment. To advance their manufacturing, real-time monitoring of adenovirus production is essential for process optimization, yield improvement, and product quality assurance. However, despite modern bioreactors being equipped with advanced sensors, comprehensive monitoring of viral production in real-time remains limited. In this work, we investigated potential indicators of the cell density effect (CDE), which imposes a trade-off between high cell density and specific productivity, thereby hindering process optimization. Glucose availability (expressed as glucose per cell) at time of infection was identified as the most effective proxy indicator. We developed an age-structured mechanistic model linking measurable process variables, such as glucose and viable cell density to viral production while capturing the CDE. This framework enables the prediction of viral particle titers in both batch and fed-batch operation modes, achieving an RMSE of 0.39 in log₁₀ IVP/mL for infectious viral particles and 0.20 in log₁₀ VP/mL for total viral particles, corresponding to approximately 2.6 and 2.0 times the measurement standard deviation of the respective assays. The proposed model and calibration method lay the foundation for a digital twin of adenovirus production, bridging real-time sensing with model-based simulation to support adaptive control and accelerate the transition toward Biopharma 4.0.