Data fusion strategies for performance improvement of a Process Analytical Technology platform consisting of four instruments: An electrospinning case study

Quantitative analysis of moisture content and particle size in a fluidized bed granulation process using near infrared spectroscopy and acoustic emission combined with data fusion strategies

Monitoring granule property is essential for fluidization maintenance and product quality control in fluidized bed granulation (FBG). In this study, two non-invasive techniques, near-infrared (NIR) spectroscopy and acoustic emission (AE), were applied for quantitative analysis of moisture content (MC) and median particle size (D50) in a FBG process, combined with chemometrics and data fusion strategies. Partial least squares (PLS) and support vector machine (SVM) regression models were established based on NIR and AE spectral data. The optimal quantitative models were identified considering the effect of spectra preprocessing and variable selection. In the comparison study, the best separate models for MC and D50 quantification were based on NIR and AE, respectively. The NIR model exhibited the better prediction ability with the determination coefficient of validation set (R2v) of 0.9815, root mean square error of validation set (RMSEv) of 0.2226 %, and residual predictive deviation (RPD) of 7.4674 for MC. Meanwhile, the AE model presented the better prediction performance with R2v of 0.9710, RMSEv of 18.2643 μm, and RPD of 5.9740 for D50. Furthermore, among three data fusion strategies, the high-level fusion model achieved the best overall performance on D50 quantification with R2v of 0.9863, RMSEv of 12.5707 μm, and RPD of 8.6798. The results indicated that both NIR and AE are effective monitoring tools for MC and D50 analysis in fluidized bed granulation process. In addition, a more accurate and reliable analysis of particle size can be achieved by combining NIR and AE technology with high-level data fusion.

Development of a PAT platform for the prediction of granule tableting properties

In this work, the feasibility of implementing a process analytical technology (PAT) platform consisting of Near Infrared Spectroscopy (NIR) and particle size distribution (PSD) analysis was evaluated for the prediction of granule downstream processability. A Design of Experiments-based calibration set was prepared using a fluid bed melt granulation process by varying the binder content, granulation time, and granulation temperature. The granule samples were characterized using PAT tools and a compaction simulator in the 100-500 kg load range. Comparing the systematic variability in NIR and PSD data, their complementarity was demonstrated by identifying joint and unique sources of variation. These particularities of the data explained some differences in the performance of individual models. Regarding the fusion of data sources, the input data structure for partial least squares (PLS) based models did not significantly impact the predictive performance, as the root mean squared error of prediction (RMSEP) values were similar. Comparing PLS and artificial neural network (ANN) models, it was observed that the ANNs systematically provided superior model performance. For example, the best tensile strength, ejection stress, and detachment stress prediction with ANN resulted in an RMSEP of 0.119, 0.256, and 0.293 as opposed to the 0.180, 0.395, and 0.430 RMSEPs of the PLS models, respectively. Finally, the robustness of the developed models was assessed.

Challenges and Opportunities of Implementing Data Fusion in Process Analytical Technology-A Review

The release of the FDA's guidance on Process Analytical Technology has motivated and supported the pharmaceutical industry to deliver consistent quality medicine by acquiring a deeper understanding of the product performance and process interplay. The technical opportunities to reach this high-level control have considerably evolved since 2004 due to the development of advanced analytical sensors and chemometric tools. However, their transfer to the highly regulated pharmaceutical sector has been limited. To this respect, data fusion strategies have been extensively applied in different sectors, such as food or chemical, to provide a more robust performance of the analytical platforms. This survey evaluates the challenges and opportunities of implementing data fusion within the PAT concept by identifying transfer opportunities from other sectors. Special attention is given to the data types available from pharmaceutical manufacturing and their compatibility with data fusion strategies. Furthermore, the integration into Pharma 4.0 is discussed.

Real-time coating thickness measurement and defect recognition of film coated tablets with machine vision and deep learning

This paper presents a system, where images acquired with a digital camera are coupled with image analysis and deep learning to identify and categorize film coating defects and to measure the film coating thickness of tablets. There were 5 different classes of defective tablets, and the YOLOv5 algorithm was utilized to recognize defects, the accuracy of the classification was 98.2%. In order to characterize coating thickness, the diameter of the tablets in pixels was measured, which was used to measure the coating thickness of the tablets. The proposed system can be easily scaled up to match the production capability of continuous film coaters. With the developed technique, the complete screening of the produced tablets can be achieved in real-time resulting in the improvement of quality control.

In-Depth Understanding of Granule Compression Behavior under Variable Raw Material and Processing Conditions

Tablet manufacturing involves the processing of raw materials through several unit operations. Thus, the mitigation of input-induced variability should also consider the downstream processability of intermediary products. The objective of the present work was to study the effect of variable raw materials and processing conditions on the compression properties of granules containing two active pharmaceutical ingredients (APIs) and microcrystalline cellulose. Differences in compressibility and tabletability of granules were highlighted in function of the initial particle size of the first API, granule polydispersity and fragmentation. Moreover, interactions were underlined with the atomizing pressure. Changing the supplier of the second API was efficiently controlled by adapting the binder addition rate and atomizing pressure during granulation, considering the starting crystal size. By fitting mathematical models on the available compression data, the influence of diluent source on granule compactibility and tabletability was identified. These differences resumed to the ease of compaction, tableting capacity and pressure sensitivity index due to variable water binding capacity of microcrystalline cellulose. Building the design space enabled the identification of suitable API types and the appropriate processing conditions (spray rate, atomizing pressure, compression force) required to ensure the desired tableting performance.

Using a material database and data fusion method to accelerate the process model development of high shear wet granulation

High shear wet granulation (HSWG) has been wildly used in manufacturing of oral solid dosage (OSD) forms, and process modeling is vital to understanding and controlling this complex process. In this paper, data fusion and multivariate modeling technique were applied to develop a formulation-process-quality model for HSWG process. The HSWG experimental data from both literature and the authors' laboratory were fused into a single and formatted representation. A material database and material matching method were used to compensate the incomplete physical characterization of literature formulation materials, and dimensionless parameters were utilized to reconstruct process variables at different granulator scales. The exploratory study on input materials properties by principal component analysis (PCA) revealed that the formulation data collected from different articles generated a formulation library which was full of diversity. In prediction of the median granule size, the partial least squares (PLS) regression models derived from literature data only and a combination of literature data and laboratory data were compared. The results demonstrated that incorporating a small number of laboratory data into the multivariate calibration model could help significantly reduce the prediction error, especially at low level of liquid to solid ratio. The proposed data fusion methodology was beneficial to scientific development of HSWG formulation and process, with potential advantages of saving both experimental time and cost.

Multivariate feed forward process control and optimization of an industrial, granulation based tablet manufacturing line using historical data

The purpose of this work was to understand the variability in disintegration time and tableting yield of high drug load (>60%) tablets prepared by batch-wise high shear wet granulation. The novelty of the study is the use of multivariate methods (Batch Evolution Models - BEMs and Batch Level Models - BLMs) to enhance process control, with a feed forward component, using prediction models built from a historical dataset acquired for 95 industrial scale batches. Time dependent process variables and significant influences on investigated parameters were identified. Prediction of output from input was tested with Partial Least Squares (PLS) and Artificial Neural Network (ANN) modeling. A reliable prediction ability was achieved for granulation water amount (±2 kg in a 16-31 kg range), tableting speed (±5000 tablets/h in a 23,000-72,500 tabl./h range) and disintegration time of cores (±100 s; in a 250-900 s range). Offsets from the optimal process evolution and certain raw material properties were correlated with differences observed in the output variables. Improvement options were identified for 80% of the batches with high disintegration time. Hence, the trained models can be applied for systematic process improvement, enabling feed forward control.

Multianalyzer Spectroscopic Data Fusion for Soil Characterization

The ability to rapidly conduct in-situ chemical analysis of multiple samples of soil and other geological materials in the field offers many advantages over a traditional approach that involves collecting samples for subsequent examination in the laboratory. This study explores the application of complementary spectroscopic analyzers and a data fusion methodology for the classification/discrimination of >100 soil samples from sites across the United States. Commercially available, handheld analyzers for X-ray fluorescence spectroscopy (XRFS), Raman spectroscopy (RS), and laser-induced breakdown spectroscopy (LIBS) were used to collect data both in the laboratory and in the field. Following a common data pre-processing protocol, principal component analysis (PCA) and partial least squares discriminant analysis (PLSDA) were used to build classification models. The features generated by PLSDA were then used in a hierarchical classification approach to assess the relative advantage of information fusion, which increased classification accuracy over any of the individual sensors from 80-91% to 94% and 64-93% to 98% for the two largest sample suites. The results show that additional testing with data sets for which classification with individual analyzers is modest might provide greater insight into the limits of data fusion for improving classification accuracy.

A mid-level data fusion approach for evaluating the internal and external changes determined by FT-NIR, electronic nose and colorimeter in Curcumae Rhizoma processing

The aim of this study was to establish a robust and rapid method for identification of the changes of Curcumae Rhizoma (CR) in its physical properties and the components after processing. Color and odor used as the indexes for physical property evaluation in this study were quantified by non-targeted bionic detectors colorimeter and e-nose according to Chinese pharmacopoeia, and all chemical changes of the samples were observed by NIR. The technique of mid-level data fusion was adopted to simulate the recognition mode of human, and the prediction accuracy was 100%. A modified NIR band extraction method was used for feature selection after removing the redundant data which otherwise may compromise the accuracy of analysis, with a result showing bisdemethoxycurcumin and curcumol as the key factors most correlated with the changes during processing. The fused matrix was analyzed to evaluate the weight of different sensors and precisely describe the external changes of the samples, followed by a pairwise correlation analysis to investigate the role of colorimeter, e-nose and NIR in identifying the changes caused by processing as well as their correlation with each other. The results of data fusion revealed that aromatic derivatives produced during processing were closely associated with the changes in external characteristics, i.e., color and smell of the samples, while the changes in proteins did not cause significant differences. Correlation analysis demonstrated that bionic sensors could be classified into two groups: a*, WW and WS sensors, which were related to NIR band at about 6500-6700 cm^-1 formed by NH vibration of amide and protein degradation, were sensitive to the processed CR; while L*, b* and WC sensors were found to be correlated to NIR band around 8000 cm^-1 which was caused by first overtone of C-H combination of aromatic derivatives, thus could be employed as the detectors for raw CR.