Blend uniformity end-point determination using near-infrared spectroscopy and multivariate calibration

Machine Learning-Enabled NIR Spectroscopy. Part 2: Workflow for Selecting a Subset of Samples from Publicly Accessible Data

An increasingly large dataset of pharmaceutics disciplines is frequently challenging to comprehend. Since machine learning needs high-quality data sets, the open-source dataset can be a place to start. This work presents a systematic method to choose representative subsamples from the existing research, along with an extensive set of quality measures and a visualization strategy. The preceding article (Muthudoss et al.. in AAPS PharmSciTech 23, 2022) describes a workflow for leveraging near infrared (NIR) spectroscopy to obtain reliable and robust data on pharmaceutical samples. This study describes the systematic and structured procedure for selecting subsamples from the historical data. We offer a wide range of in-depth quality measures, diagnostic tools, and visualization techniques. A real-world, well-researched NIR dataset was employed to demonstrate this approach. This open-source tablet dataset ( http://www.models.life.ku.dk/Tablets ) consists of different doses in milligrams, different shapes, and sizes of dosage forms, slots in tablets, three different manufacturing scales (lab, pilot, production), coating differences (coated vs uncoated), etc. This sample is appropriate; that is, the model was developed on one scale (in this research, the lab scale), and it can be great to investigate how well the top models are transferable when tested on new data like pilot-scale or production (full) scale. A literature review indicated that the PLS regression models outperform artificial neural network-multilayer perceptron (ANN-MLP). This work demonstrates the selection of appropriate hyperparameters and their impact on ANN-MLP model performance. The hyperparameter tuning approaches and performance with available references are discussed for the data under investigation. Model extension from lab-scale to pilot-scale/production scale is demonstrated. HIGHLIGHTS: • We present a comprehensive quality metrics and visualization strategy in selecting subsamples from the existing studies • A comprehensive assessment and workflow are demonstrated using historical real-world near-infrared (NIR) data sets • Selection of appropriate hyperparameters and their impact on artificial neural network-multilayer perceptron (ANN-MLP) model performance • The choice of hyperparameter tuning approaches and performance with available references are discussed for the data under investigation • Model extension from lab-scale to pilot-scale successfully demonstrated.

Machine Learning-Enabled NIR Spectroscopy in Assessing Powder Blend Uniformity: Clear-Up Disparities and Biases Induced by Physical Artefacts

NIR spectroscopy is a non-destructive characterization tool for the blend uniformity (BU) assessment. However, NIR spectra of powder blends often contain overlapping physical and chemical information of the samples. Deconvoluting the information related to chemical properties from that associated with the physical effects is one of the major objectives of this work. We achieve this aim in two ways. Firstly, we identified various sources of variability that might affect the BU results. Secondly, we leverage the machine learning-based sophisticated data analytics processes. To accomplish the aforementioned objectives, calibration samples of amlodipine as an active pharmaceutical ingredient (API) with the concentrations ranging between 67 and 133% w/w (dose ~ 3.6% w/w), in powder blends containing excipients, were prepared using a gravimetric approach and assessed using NIR spectroscopic analysis, followed by HPLC measurements. The bias in NIR results was investigated by employing data quality metrics (DQM) and bias-variance decomposition (BVD). To overcome the bias, the clustered regression (non-parametric and linear) was applied. We assessed the model's performance by employing the hold-out and k-fold internal cross-validation (CV). NIR-based blend homogeneity with low mean absolute error and an interval estimates of 0.674 (mean) ± 0.218 (standard deviation) w/w was established. Additionally, bootstrapping-based CV was leveraged as part of the NIR method lifecycle management that demonstrated the mean absolute error (MAE) of BU ± 3.5% w/w and BU ± 1.5% w/w for model generalizability and model transferability, respectively. A workflow integrating machine learning to NIR spectral analysis was established and implemented. Impact of various data learning approaches on NIR spectral data.

Real-time monitoring of the column chromatographic process of Phellodendri Chinensis Cortex part II: multivariate statistical process control based on near-infrared spectroscopy

Multivariate statistical process control has been successfully used for the real-time monitoring of the column chromatographic process of Phellodendri Chinensis Cortex.

Pharmaceutical application of multivariate modelling techniques: a review on the manufacturing of tablets

The tablet manufacturing process is a complex system, especially in continuous manufacturing (CM). It includes multiple unit operations, such as mixing, granulation, and tableting. In tablet manufacturing, critical quality attributes are influenced by multiple factorial relationships between material properties, process variables, and interactions. Moreover, the variation in raw material attributes and manufacturing processes is an inherent characteristic and seriously affects the quality of pharmaceutical products. To deepen our understanding of the tablet manufacturing process, multivariable modeling techniques can replace univariate analysis to investigate tablet manufacturing. In this review, the roles of the most prominent multivariate modeling techniques in the tablet manufacturing process are discussed. The review mainly focuses on applying multivariate modeling techniques to process understanding, optimization, process monitoring, and process control within multiple unit operations. To minimize the errors in the process of modeling, good modeling practice (GMoP) was introduced into the pharmaceutical process. Furthermore, current progress in the continuous manufacturing of tablets and the role of multivariate modeling techniques in continuous manufacturing are introduced. In this review, information is provided to both researchers and manufacturers to improve tablet quality.

Near infrared analysis of pharmaceutical powders with empirical target distribution optimization (ETDO)

Near infrared (NIR) spectroscopy is a well-established method for analysis of pharmaceutical products, and especially useful for process monitoring and control of continuous production due to high sample throughput. In this work, a previously established method called empirical target distribution optimization (ETDO) wherein reference sample values using information from model prediction of the calibration data was used as a tool to improve the performance of NIR partial least squares (PLS) models. Model performance was assessed using root mean square error (R²), bias and accuracy in prediction of test samples. A target value selection threshold was tested to assess the ETDO procedure for NIR analysis of powder samples. The amount of specific variation captured by the model was examined and compared for models calibrated with and without ETDO. The results reported in this work suggests that PLS models optimized with ETDO of reference values can provide more specific PLS models for NIR analysis for complex powder mixtures. In addition, the model optimization method could also be applied as a tool to verify the necessary amount of PLS components to produce robust models. The ETDO method presented in this work is an approach that could be applied in the development of continuous blending or tableting processes where robust in-line quantitative analysis of powder samples is needed.

Methods to Assess Mixing of Pharmaceutical Powders

The pharmaceutical manufacturing process consists of several steps, each of which must be monitored and controlled to ensure quality standards are met. The level of blending has an impact on the final product quality; therefore, it is important to be able to monitor blending progress and identify an end-point. Currently, the pharmaceutical industry assesses blend content and uniformity through the extraction of samples using thief probes followed by analytical methods, such as spectroscopy, to determine the sample composition. The development of process analytical technologies (PAT) can improve product monitoring with the aim of increasing efficiency, product quality and consistency, and creating a better understanding of the manufacturing process. Ideally, these are inline methods to remove issues related to extractive sampling and allow direct monitoring of the system using various sensors. Many technologies have been investigated, including spectroscopic techniques such as near-infrared spectroscopy, velocimetric techniques that may use tracers, tomographic techniques, and acoustic emissions monitoring. While some techniques have demonstrated potential, many have significant disadvantages including the need for equipment modification, specific requirements of the material, expensive equipment, extensive analysis, the location of the probes may be critical and/or invasive, and lastly, the technique may only be applicable to the development phase. Both the advantages and disadvantages of the technologies should be considered in application to a specific system.

PAT-Based Control of Fluid Bed Coating Process Using NIR Spectroscopy to Monitor the Cellulose Coating on Pharmaceutical Pellets

Current endeavor was aimed towards monitoring percent weight build-up during functional coating process on drug-layered pellets. Near-infrared (NIR) spectroscopy is an emerging process analytical technology (PAT) tool which was employed here within quality by design (QbD) framework. Samples were withdrawn after spraying every 15-Kg cellulosic coating material during Wurster coating process of drug-loaded pellets. NIR spectra of these samples were acquired using cup spinner assembly of Thermoscientific Antaris II, followed by multivariate analysis using partial least squares (PLS) calibration model. PLS model was built by selecting various absorption regions of NIR spectra for Ethyl cellulose, drug and correlating the absorption values with actual percent weight build up determined by HPLC. The spectral regions of 8971.04 to 8250.77 cm^-1, 7515.24 to 7108.33 cm^-1, and 5257.00 to 5098.87 cm^-1 were found to be specific to cellulose, where as the spectral region of 6004.45 to 5844.14 cm^-1was found to be specific to drug. The final model gave superb correlation co-efficient value of 0.9994 for calibration and 0.9984 for validation with low root mean square of error (RMSE) values of 0.147 for calibration and 0.371 for validation using 6 factors. The developed correlation between the NIR spectra and cellulose content is useful in precise at-line prediction of functional coat value and can be used for monitoring the Wurster coating process.

Direct compaction: An update of materials, trouble-shooting, and application

Direct compaction (DC) is the preferred choice for tablet manufacturing; however, only less than 20% of active pharmaceutical ingredients could be compacted via DC as its high requirement for functional properties of materials. Materials with improper functionalities could lead to serious troubles during DC manufacturing, such as content non-uniformity, sticking, and capping, all of which profoundly affect the properties of final products and, thus, severely restrict the practical application of DC. With undoubted importance, these seem to be unexpectedly ignored by reviewers but not researchers in terms of many original research articles published recently. Therefore, as an informative supplement and update, this review mainly focused on trouble-shooting and application situation of DC, together with several newly reported materials.

Real-time imaging as an emerging process analytical technology tool for monitoring of fluid bed coating process

A direct imaging system (Eyecon^TM) was used as a Process Analytical Technology (PAT) tool to monitor fluid bed coating process. Eyecon^TM generated real-time onscreen images, particle size and shape information of two identically manufactured laboratory-scale batches. Eyecon^TM has accuracy of measuring the particle size increase of ±1 μm on particles in the size range of 50-3000 μm. Eyecon^TM captured data every 2 s during the entire process. The moving average of D90 particle size values recorded by Eyecon^TM were calculated for every 30 min to calculate the radial coating thickness of coated particles. After the completion of coating process, the radial coating thickness was found to be 11.3 and 9.11 μm, with a standard deviation of ±0.68 and 1.8 μm for Batch 1 and Batch 2, respectively. The coating thickness was also correlated with percent weight build-up by gel permeation chromatography (GPC) and dissolution. GPC indicated weight build-up of 10.6% and 9.27% for Batch 1 and Batch 2, respectively. In conclusion, weight build-up of 10% can also be correlated with 10 ± 2 μm increase in the coating thickness of pellets, indicating the potential applicability of real-time imaging as an endpoint determination tool for fluid bed coating process.

Blend uniformity evaluation during continuous mixing in a twin screw granulator by in-line NIR using a moving F-test

This study focuses on the twin screw granulator of a continuous from-powder-to-tablet production line. Whereas powder dosing into the granulation unit is possible from a container of preblended material, a truly continuous process uses several feeders (each one dosing an individual ingredient) and relies on a continuous blending step prior to granulation. The aim of the current study was to investigate the in-line blending capacity of this twin screw granulator, equipped with conveying elements only. The feasibility of in-line NIR (SentroPAT, Sentronic GmbH, Dresden, Germany) spectroscopy for evaluating the blend uniformity of powders after the granulator was tested. Anhydrous theophylline was used as a tracer molecule and was blended with lactose monohydrate. Theophylline and lactose were both fed from a different feeder into the twin screw granulator barrel. Both homogeneous mixtures and mixing experiments with induced errors were investigated. The in-line spectroscopic analyses showed that the twin screw granulator is a useful tool for in-line blending in different conditions. The blend homogeneity was evaluated by means of a novel statistical method being the moving F-test method in which the variance between two blocks of collected NIR spectra is evaluated. The α- and β-error of the moving F-test are controlled by using the appropriate block size of spectra. The moving F-test method showed to be an appropriate calibration and maintenance free method for blend homogeneity evaluation during continuous mixing.

Use of near-infrared spectroscopy and multipoint measurements for quality control of pharmaceutical drug products

Near-infrared (NIR) spectroscopy is a non-destructive analytical technique that enables better-understanding and optimization of pharmaceutical processes and final drug products. The use in line is often limited by acquisition speed and sampling area. This work focuses on performing a multipoint measurement at high acquisition speed at the end of the manufacturing process on a conveyor belt system to control both the distribution and the content of active pharmaceutical ingredient within final drug products, i.e., tablets. A specially designed probe with several collection fibers was developed for this study. By measuring spectral and spatial information, it provides physical and chemical knowledge on the final drug product. The NIR probe was installed on a conveyor belt system that enables the analysis of a lot of tablets. The use of these NIR multipoint measurement probes on a conveyor belt system provided an innovative method that has the potential to be used as a new paradigm to ensure the drug product quality at the end of the manufacturing process and as a new analytical method for the real-time release control strategy. Graphical abstract Use of near-infrared spectroscopy and multipoint measurements for quality control of pharmaceutical drug products.

Detecting Blending End-Point Using Mean Squares Successive Difference Test and Near-Infrared Spectroscopy

An algorithm based on mean squares successive difference test applied to near-infrared and principal component analysis scores was developed to monitor and determine the blending profile and to assess the end-point in the statistical stabile phase. Model formulations consisting of an active compound (acetylsalicylic acid), together with microcrystalline cellulose and two grades of calcium carbonate with dramatically different particle shapes, were prepared. The formulation comprising angular-shaped calcium carbonate reached blending end-point slower when compared with the formulation comprising equant-shaped calcium carbonate. Utilizing the ring shear test, this distinction in end-point could be related to the difference in flowability of the formulations. On the basis of the two model formulations, a design of experiments was conducted to characterize the blending process by studying the effect of CaCO3 grades and fill level of the bin on blending end-point. Calcium carbonate grades, fill level, and their interaction were shown to have a significant impact on the blending process.

A comparative study of using in-line near-infrared spectra, ultraviolet spectra and fused spectra to monitor Panax notoginseng adsorption process

The step of enriching and purifying saponins by macroporous resin column chromatography is closely related to the safety and efficacy of Panax notoginseng products during their manufacturing processes. Adsorption process is one of the most critical unit operations within each chromatographic cycle. In order to understand the adsorption process directly, it is necessary to develop a rapid and precise method to monitor the adsorption process in real time. In this study, comparative evaluation of using near-infrared (NIR) spectra, ultraviolet (UV) spectra and fused spectra to monitor the adsorption process of P. notoginseng was conducted. The uninformative variable elimination by partial least squares (UVE-PLS) regression models were established for quantification of notoginsenoside R1, ginsenoside Rg1, ginsenoside Re, ginsenoside Rb1 and ginsenoside Rd in effluents based on different spectra. There was a significant improvement provided by the models based on fused spectra. The results in this work were conducive to solving the problems about real-time quantitative analysis of saponins during P. notoginseng adsorption. The fusion method of NIR and UV spectra combined with UVE-PLS regression could be a promising strategy to real-time analyze the components, which are difficult to be quantified by individual spectroscopic technique.

Verification of model development technique for NIR-based real-time monitoring of ingredient concentration during blending

There has been a considerable research on the process analytical technology (PAT) and real-time monitoring based on NIR, but the model development is still an important issue and persons in charge have difficulty in building good models. In this study, to realize efficient NIR-based real-time monitoring of ingredient concentration and establish a model development method, we investigated the effect of a calibration set, spectral preprocessing, wavelengths, and other factors on the prediction error through pilot and commercial scale blending experiments. The results confirmed that the small prediction error was realized by a calibration set, including dynamic measurement spectra acquired with the target blender. In addition, the results demonstrated that locally weighted partial least squares (LW-PLS) achieved the smaller prediction error than conventional PLS. The present study has also clarified that spectral preprocessing methods and wavelengths selected to build a model affect the prediction error of ingredient concentration interactively. A wide wavelength range should be selected when the spectral preprocessing does not lessen the effect of baseline variation, while a narrow wavelength range should be selected when it strongly decreases the effect.

High-throughput NIR-chemometric methods for determination of drug content and pharmaceutical properties of indapamide tablets

This paper describes the development, validation and application of NIR-chemometric methods for API content and pharmaceutical characterization (disintegration time and crushing strength) of indapamide intact tablets. Development of the method for chemical characterization was performed on samples corresponding to 80, 90, 100, 110 and 120% of indapamide content and for pharmaceutical characterization on samples prepared at nine different compression forces (covering the interval 7-45 kN). NIR spectra of prepared tablets were recorded in transmission mode, and partial least-squares followed by leave-one-out cross-validation were used to develop models for the prediction of the drug content and the pharmaceutical properties of tablets. All developed models were validated in terms of trueness, precision and accuracy. No statistical differences were found between results predicted by NIR-chemometric methods and the ones determined by reference methods. Therefore, the developed NIR-chemometric methods meet the requirements of a high-throughput method for the determination of drug content, pharmaceutical properties of indapamide tablets.