A multi-step machine learning approach for accelerating QbD-based process development of protein spray drying.

This study proposes a new material-efficient multi-step machine learning (ML) approach for the development of a design space (DS) for spray drying proteins. Typically, a DS is developed by performing a design of experiments (DoE) with the spray dryer and the protein of interest, followed by deriving the DoE models via multi-variate regression. This approach was followed as a benchmark to the ML approach. The more complex the process and required accuracy of the final model is, the more experiments are necessary. However, most biologics are expensive and thus experiments should be kept to a minimum. Therefore, the suitability of using a surrogate material and ML for the development of a DS was investigated. To this end, a DoE was performed with the surrogate and the data used for training the ML approach. The ML and DoE model predictions were compared to measurements of three protein-based validation runs. The suitability of using lactose as surrogate was investigated and advantages of the proposed approach were demonstrated. Limitations were identified at protein concentrations >35 mg/ml and particle sizes of x50>6 µm. Within the investigated DS protein secondary structure was preserved, and most process settings, resulted in yields >75% and residual moisture <10 wt%.

Steady-state modeling of a new continuous API dryer: Reduced-order model and investigation of dryer performance.

In the present study, a reduced-order model is proposed to analyze a novel continuous dryer with an application in the pharmaceutical industry. The model was validated using process data from ibuprofen drying test runs, and the results were in good agreement with the experimental data. The test substance was an ibuprofen paste with an initial LOD of up to 30 w%. The simulations showed that the contact heat transfer coefficient can be correlated with the degree of wetness. Furthermore, a set of simulations was performed to analyze the influence of input parameters on the dryer's performance: i) the inlet air flow rate and ii) the inlet air temperature. The simulation results demonstrated that a variation in the inlet air temperature significantly affects the air temperature profile, while the inlet air flow rate has a minor effect. Besides, it was also established that the inlet solid LoD has the most considerable effect on the product quality (e.g., final solid moisture content). The results showed a deviation of less than 10% for the product LoD and the product temperature in most cases.

RTD modeling of a continuous dry granulation process for process control and materials diversion.

Disturbance propagation during continuous manufacturing processes can be predicted by evaluating the residence time distribution (RTD) of the specific unit operations. In this work, a dry granulation process was modelled and four scenarios of feeding events were simulated. We performed characterization of the feeders and developed RTD models for the blender and the roller compactor based on impulse-response measurements via color tracers. Out-of-specification material was defined based on the active pharmaceutical ingredient (API) concentration. We calculated the amount of waste material at various diversion points, considering four feeder-related process-upset scenarios and formulated considerations for the development of a control concept. The developed RTD models allow material tracking of materials that may be used for following the spread contaminants within the process and for batch definition. The results show that RTD modeling is a valuable tool for process development and design, as well as for process monitoring and material tracking.