Publications

Authors: I. Muthancheri, R. Ramachandran

Paper Link: Link

Abstract: In this study, a hybrid modeling framework was developed for predicting size distribution and content uniformity of granules in a bi-component wet granulation system with components of differing hydrophobicities. Two bi-component formulations, (1) ibuprofen-USP and micro-crystalline cellulose and (2) micronized acetaminophen and micro-crystalline cellulose, were used in this study. First, a random forest method was used for predicting the probability of nucleation mechanism (immersion and solid spread), depending upon the formulation hydrophobicity. The predicted nucleation mechanism probability is used to determine the aggregation rate as well as the initial particle distribution in the population balance model. The aggregation process was modeled as Type-I: Sticking aggregation and Type-II: Deformation-driven aggregation. In Type-I, the capillary force dominant aggregation mechanism is represented by the particles sticking together without deformation. In the case of Type-II, the particle deformation causes an increase in the contact area, representing a viscous force dominant aggregation mechanism. The choice between Type-I and II aggregation is determined based on the difference in nucleation mechanism that is predicted using the random forest method. The model was optimized and validated using the granule content uniformity data and size distribution data obtained from the experimental studies. The proposed framework predicted content non-uniform behavior for formulations that favored immersion nucleation and uniform behavior for formulations that favored solid-spreading nucleation.

Authors: S. Gupta, Y. Baranwal, A.D Roman-Ospino, D. Hausner, R. Ramachandran, F.J. Muzzio

Paper Link: Link

Abstract: Raman chemical mapping is an inherently slow analysis tool. Accurate and robust multivariate analysis algorithms, which require least amount of time and effort in method development are desirable. Calibration-free regression and resolution approaches such as classical least squares (CLS) and multivariate curve resolution using alternating least squares (MCR-ALS), respectively, help in reducing the resources required for method development. However, conventional CLS does not consider appropriate constraints, which may result in negative and/or greater than 100 % Raman concentration scores, while MCR-ALS may not always be as accurate as regression-based algorithms. Linear iterative optimization technology (IOT) is another calibration-free algorithm, which with appropriate constraints has previously shown promise in online and offline pharmaceutical mixture composition determination. This paper aims to evaluate the performance of the linear IOT algorithm for Raman chemical mapping of the active pharmaceutical ingredient (API), diluent, and lubricant in pharmaceutical tablets. Two pre-processing strategies were applied to the raw Raman mapping spectra. The results were compared with CLS (current reference method) and MCR-ALS. Special emphasis was given to mapping at low Raman exposure times to enable feasible total acquisition times (< 5 h). The quality of IOT/CLS/MCR-ALS estimated Raman concentration predictions were assessed by calculating a correlation factor between the spectrum corresponding to the maximum predicted concentration (or resolved spectra) of a component for IOT/CLS (or MCR-ALS) and the pure powder component spectrum. The Raman chemical maps were visualized, and the average Raman concentrations scores were compared. The results demonstrated the utility of IOT in Raman chemical mapping of pharmaceutical tablets. The diluent (lactose) and API (semi-fine APAP) used in this study were reliably estimated by IOT at relatively short Raman exposure times. On the other hand, as expected, the lubricant (magnesium stearate) could not be detected in any of the cases investigated here, irrespective of the algorithm used. Overall, for the API and diluent used in this formulation as well as the chemical mapping conditions, linear IOT seemed to better estimate the pure spectrum intensities and the average Raman scores (closer to CLS) in comparison to MCR-ALS. Moreover, application of appropriate constraints in linear IOT avoided the presence of negative and/or greater than 100 % Raman concentration scores, as observed in CLS-based Raman chemical maps.

Authors: L. Kotamarthy, R. Ramachandran

Paper Link: Link

Abstract: Wet granulation is a size enlargement process in which fine powders are agglomerated into larger granules via the presence of a liquid binder. Among various types and modes of granulation processes, twin-screw granulation is a popular continuous wet granulation technique. The typical residence time that powders spend in a twin-screw granulator are in the order of seconds, hence, the rigorous mixing of solid and liquid particles across particle sizes and equipment geometry becomes even more important. This study aims to improve the mechanistic understanding of the effect of process and screw parameters on the mixing dynamics inside a twin-screw granulator using dispersion coefficient as a key metric. The dispersion coefficient quantifies the dispersion and is directly proportional to the extent of mixing occurring inside the granulator. An analysis of the experimentally obtained dispersion coefficient is performed to understand the effect of key process and design parameters such as feed rate, screw speed, number of kneading elements and, stagger angle. A semi-mechanistic model that incorporates these parameters was developed to estimate and predict the dispersion coefficient. The model accurately predicts the dispersion coefficient values and the goodness-of-fit values (R2) for the test set and full data set were found to be equal to 0.920 and 0.932 respectively. This model was further tested by predicting the complete RTD curves using the predicted dispersion coefficients. The average goodness-of-fit value (R2) for the prediction of RTD curves for all the experimental runs was calculated to be equal to 0.698.

Authors: I. Muthancheri, S. Oka, R. Ramachandran

Paper Link: Link

Abstract: In this study, the effect of hydrophobic components was analyzed through single drop nucleation experiments on a particle bed. The particle bed’s relative hydrophobicity/wettability was varied by changing the percentage composition of hydrophobic and hydrophilic glass beads. The generated nuclei were then isolated and tested using near-infrared spectroscopy and image analysis. The data collected from repeating the experiments were then processed using a random forest machine learning model to predict the probability of immersion and solid-spread nucleation. This model was used to indicate the nucleation mechanism of two formulations. The first contains a varying composition of Ibuprofen and Microcrystalline cellulose, and the second consists of varying compositions of Acetaminophen and Microcrystalline cellulose. The model was validated with experiments. At higher percentage composition of Acetaminophen, the model predicted a higher probability of immersion nucleation. At a higher concentration of Ibuprofen, the model predicted a high likelihood of solid-spread nucleation.

Authors: C. Sampat, R. Ramachandran

Paper Link: Link

Abstract: The digitization of manufacturing processes has led to an increase in the availability of process data, which has enabled the use of data-driven models to predict the outcomes of these manufacturing processes. Data-driven models are instantaneous in simulate and can provide real-time predictions but lack any governing physics within their framework. When process data deviates from original conditions, the predictions from these models may not agree with physical boundaries. In such cases, the use of first-principle-based models to predict process outcomes have proven to be effective but computationally inefficient and cannot be solved in real time. Thus, there remains a need to develop efficient data-driven models with a physical understanding about the process. In this work, we have demonstrate the addition of physics-based boundary conditions constraints to a neural network to improve its predictability for granule density and granule size distribution (GSD) for a high shear granulation process. The physics-constrained neural network (PCNN) was better at predicting granule growth regimes when compared to other neural networks with no physical constraints. When input data that violated physics-based boundaries was provided, the PCNN identified these points more accurately compared to other non-physics constrained neural networks, with an error of <1%. A sensitivity analysis of the PCNN to the input variables was also performed to understand individual effects on the final outputs.

Authors: I. Muthancheri, A. Chaturbedi, A. Betard, R. Ramachandran.

Paper Link: Link

Abstract: This paper presents a predictive modeling approach of the high shear wet granulation process, quantifying the difference between the steady and induction granule growth behavior. The spatial heterogeneity in liquid binder distribution and shear rate is simulated using a compartmental population balance model. The granulator is divided into two compartments based on particle motion, which consists of a circulation compartment, and an impeller compartment. In the circulation compartment, a viscous dissipation dependent coalescence kernel is adapted for the aggregation process. In the impeller compartment a shear rate dependent aggregation kernel is implemented. The model was calibrated and validated using the dynamic evolution of granule mean size (d50). The granulation dynamics are studied with respect to change in impeller speed, liquid to solid ratio, wet massing time, initial porosity, and binder viscosity. The transition from induction growth to steady growth regime with changing process conditions is demonstrated using the model. It is observed that the model captures the effect of process parameters and spatial heterogeneity on the dynamic evolution of d50.

Authors: V. Chopda, A. Gyrogypal, O. Yang, R. Singh, R. Ramachandran, H. Zhang, G. Tsilomelekis, S. Chundawat, M. Ierapetritou

Paper Link: Link

Abstract: Continuous bioprocessing is significantly changing the biological drugs (or biologics) manufacturing landscape by potentially improving product quality, process stability, and overall profitability, as was similarly seen during the adoption of advanced manufacturing processes for small molecule drugs in the past decade. However, the implementation of continuous manufacturing for biological processes producing protein-based drug molecules, such as monoclonal antibodies (mAbs), is facing several new hurdles. The barriers to continuous bioprocessing can be overcome through improved process understanding via better predictive capabilities enabled by hybrid modeling that can also lead to robust process control. This review article summarizes the recent advances and ongoing obstacles faced during the use of advanced process analytical technologies (PAT), process modeling, and control strategies to enable continuous manufacturing of mAbs. In addition, this review also discusses the process strategies and future directions of advanced continuous manufacturing approaches that have been adapted by other industries and that could be implemented for mAbs production soon.

Authors: A. Roman-Ospino, Y. Baranwal, J. Li, J. Vargas, B. Igne, S. Bate, D. Brouckaert, F. Chauchard, D. Hausner, R. Ramachandran, R. Singh, F. Muzzio

Paper Link: Link

Abstract: In-line measurements of low dose blends in the feed frame of a tablet press were performed for API concentration levels as low as 0.10% w/w. The proposed methodology utilizes the advanced sampling capabilities of a Spatially Resolved Near-Infrared (SR-NIR) probe to develop Partial Least-Squares calibration models. The fast acquisition speed of multipoint spectra allowed the evaluation of different numbers of co-adds and feed frame paddle speeds to establish the optimum conditions of data collection to predict low potency blends. The interaction of the feed frame paddles with the SR-NIR probe was captured with high resolution and allowed the implementation of a spectral data selection criterion to remove the effect of the paddles from the calibration and testing process. The method demonstrated accuracy and robustness when predicting drug concentrations across different feed frame paddle speeds.

Authors: S. Muddu, L. Kotamarthy, Rohit R.

Paper Link: Link

Abstract: This work is concerned with the semi-mechanistic prediction of residence time metrics using historical data from mono-component twin screw wet granulation processes. From the data, several key parameters such as powder throughput rate, shafts rotation speed, liquid binder feed ratio, number of kneading elements in the shafts and the stagger angle between the kneading elements were identified and physical factors were developed to translate those varying parameters into expressions affecting the key intermediate phenomena in the equipment, holdup, flow and mixing. The developed relations were then tested across datasets to evaluate the performance of the model, applying a k-fold optimization technique. The semi-mechanistic predictions were evaluated both qualitatively through the main effects plots and quantitatively through the parity plots and correlations between the tuning constants across datasets. The root mean square error (RMSE) was used as a metric to compare the degree of goodness of fit for different datasets using the developed semi-mechanistic relations. In summary this paper presents a new approach at estimating both the residence time metrics in twin screw wet granulation, mean residence time (MRT) and variance through semi-mechanistic relations, the validity of which have been tested for different datasets.

Authors: K. M. Moroney, L. Kotamarthy, I. Muthancheri, R. Ramachandran, M. Vynnycky

Paper Link: Link

Abstract: Accurate mechanistic in vitro dissolution models can deliver insight into drug release behaviour and guide formulation development. Drug release profiles from drug-excipient granules can be impacted by variation of porosity and drug load within granules, which may arise from inherent variability in granulation processes. Here, we analyse and validate a recent model of drug release from a single spherical granule with a matrix of insoluble excipient, incorporating radial variation of porosity and drug load. The model is presented and specialised to the case where the initial drug load is large compared to the capacity of the granule’s pores at solubility. In this limit, the model reduces to a single ordinary differential equation describing depletion of a shrinking, drug-saturated core. Model validation is performed using drug release data from the literature for a granule system consisting of acetaminophen and microcrystalline cellulose. A new extended model to describe dissolution from a polydisperse collection of granules is derived. The performance is compared to single particle models using equivalent spherical diameters. The developed model provides a new tool to explore the dissolution parameter space for these systems and for considering the impact of radial variation of granule porosity and drug load arising from manufacturing processes.