Evaluation of Infrared-Reflection Absorption Spectroscopy Measurement and Locally Weighted Partial Least-Squares for Rapid Analysis of Residual Drug Substances in Cleaning Processes

Just-in-Time Learning-Integrated Partial Least-Squares Strategy for Accurately Predicting 71 Chemical Constituents in Chinese Tobacco by Near-Infrared Spectroscopy

Near-infrared spectroscopy has been widely used to characterize the chemical composition of tobacco because it is fast, economical, and nondestructive. However, few predictive models perform ideally when applied to large spectral libraries of tobacco and its various chemical indicators. In this study, the just-in-time learning-integrated partial least-squares (JIT-PLS) modeling strategy was applied for the first time to quantitatively analyze 71 chemical components in Chinese tobacco. Approximately 18000 tobacco samples from China were analyzed to find appropriately similar measurements and propose suitable and flexible similar subsets from the calibration for each test sample. In total, 879 representative aged tobacco leaf samples and 816 cigarette samples were used as external instances to evaluate the practical predicting ability of the proposed method. The most suitable similar subsets for each test sample could be selected by limiting the Euclidean distance and number of similar subsets to 0-3.0 × 10^-9 and 10-300, respectively. The majority of the JIT-PLS models performed significantly better than traditional PLS models. Specifically, using JIT-PLS instead of traditional PLS models increased the R ² values from 0.347-0.984 to 0.763-0.996, and from 0.179-0.981 to 0.506-0.989 for the prediction of 67 and 71 components in aged tobacco leaf and cigarette samples, respectively. Good prediction ability was demonstrated for routine chemical components, polyphenolic compounds, organic acids, and other compounds, with the mean ratios of prediction to deviation (RPD_mean) being 7.74, 4.39, 4.05, and 5.48, respectively). The proposed methodology could simultaneously determine 67 major components in large and complicated tobacco spectral libraries with high precision and accuracy, which will assist tobacco and cigarette quality control in collecting as well as processing stages.

Spectroscopy-Based Local Modeling Method for High-Throughput Quantification of Nucleic Acid Loading in Lipid Nanoparticles

Lipid nanoparticles (LNPs) are the most widely investigated delivery systems for nucleic acid-based therapeutics and vaccines. Loading efficiency of nucleic acids may vary with formulation conditions, and it is considered one of the critical quality attributes of LNP products. Current analytical methods for quantification of cargo loading in LNPs often require external standard preparations and preseparation of unloaded nucleic acids from LNPs; therefore, they are subject to tedious and lengthy procedures, LNP stability, and unpredictable recovery rates of the separated analytes. Here, we developed a modeling approach, which was based on locally weighted regression (LWR) of ultraviolet (UV) spectra of unpurified samples, to quantify the loading of nucleic acid cargos in LNPs in-situ. We trained the model to automatically tune the training library space according to the spectral features of a query sample so as to robustly predict the nucleic acid cargo concentration and rank loading capacity with similar performance as the more complicated experimental approaches. Furthermore, we successfully applied the model to a wide range of nucleic acid cargo species, including antisense oligonucleotides, single-guided RNA, and messenger RNA, in varied lipid matrices. The LWR modeling approach significantly saved analytical time and efforts by facile UV scans of 96-well sample plates within a few minutes and with minimal sample preprocessing. Our proof-of-concept study presented the very first data mining and modeling strategy to quantify nucleic acid loading in LNPs and is expected to better serve high-throughput screening workflows, thereby facilitates early-stage optimization and development of LNP formulations.

Prediction Performance and Economic Efficiency of Soft Sensors for in-Line Water Content Monitoring in Fluidized Bed Granulation: PP-Based Model vs. NIRS-Based Model

Soft sensors play a crucial role as process analytical technology (PAT) tools. They are classified into physical models, statistical models, and their hybrid models. In general, statistical models are better estimators than physical models. In this study, two types of standard statistical models using process parameters (PPs) and near-infrared spectroscopy (NIRS) were investigated in terms of prediction accuracy and development cost. Locally weighted partial least squares regression (LW-PLSR), a type of nonlinear regression method, was utilized. Development cost was defined as the cost of goods required to construct an accurate model of commercial-scale equipment. Eleven granulation lots consisting of three laboratory-scale, two pilot-scale, and six commercial-scale lots were prepared. Three commercial-scale granulation lots were selected as a validation dataset, and the remaining eight granulation lots were utilized as calibration datasets. The results demonstrated that the PP-based and NIRS-based LW-PLSR models achieved high prediction accuracy without using the commercial-scale data in the calibration dataset. This practical case study clarified that the construction of accurate LW-PLSR models requires the calibration samples with the following two features: 1) located near the validation samples on the subspace spanned by principal components (PCs), and 2) having a wide range of variations in PC scores. In addition, it was confirmed that the reduction in cost and mass fraction of active pharmaceutical ingredient (API) made the PP-based models more cost-effective than the NIRS-based models. The present work supports to build accurate models efficiently and save the development cost of PAT.

Scale-Free Soft Sensor for Monitoring of Water Content in Fluid Bed Granulation Process

In-line monitoring of granule water content during fluid bed granulation is important to control drug product qualities. In this study, a practical scale-free soft sensor to predict water content was proposed to cope with the manufacturing scale changes in drug product development. The proposed method exploits two key ideas to construct a scale-free soft sensor. First, to accommodate the changes in the manufacturing scale, the process parameters (PPs) that are critical to water content at different manufacturing scales were selected as input variables. Second, to construct an accurate statistical model, locally weighted partial least squares regression (LW-PLSR), which can cope with collinearity and nonlinearity, was utilized. The soft sensor was developed using both laboratory (approx. 4 kg) data and pilot (approx. 25 kg) scale data, and the prediction accuracy in the commercial (approx. 100 kg) scale was evaluated based on the assumption that the process was scaled-up from the pilot scale to the commercial scale. The developed soft sensor exhibited a high prediction accuracy, which was equivalent to the commonly used near-infrared (NIR) spectra-based method. The proposed method requires only standard instruments; therefore, it is expected to be a cost-effective alternative to the NIR spectra-based method.

Investigation of Preprocessing and Validation Methodologies for PAT: Case Study of the Granulation and Coating Steps for the Manufacturing of Ethenzamide Tablets

After the Food and Drug Association in the USA published guidelines on the enhanced use of process analytical technology (PAT) and continuous manufacturing, many studies regarding PAT and continuous manufacturing have been published. This paper describes a case study involving granulation and coating steps with ethenzamide to investigate interference for PAT model construction and model management. We investigated what factors should be considered and addressed when PAT is implemented for continuous manufacturing and how predictive models should be constructed. The product qualities that were monitored were moisture content and particle size in the granulation step and tablet weight and moisture content in the coating step. We have constructed models for the granulation step and validated the predictive capability of the models against an external dataset. A partial least squares (PLS) model with manual wavelength selection had the best predictive accuracy for loss on drying against the external validation set. We found that the prediction of loss on drying was accurate, but the prediction of particle size was not sufficiently accurate. In the coating step, because of the small amount of data, we performed three-fold cross-validation and y-scrambling 10 times, to select the optimal hyper-parameters and to check if the models were fitted to chance correlations. We confirmed that the coating agent weights, tablet weights, and water content could be accurately predicted based on the mean of the R² score for cross-validation. Addition of other variables, as well as the absorbance, slightly improved the predictive accuracy.

Hyperspectral imaging and improved feature variable selection for automated determination of deoxynivalenol in various genetic lines of barley kernels for resistance screening

Fusarium head blight (FHB), a fungus disease of small grain cereal crops, results in reduced yields and diminished value of harvested grain due to the presence of deoxynivalenol (DON), a mycotoxin produced by the causal pathogen Fusarium graminearum. DON and other tricothecene mycotoxins pose serious health risks to both humans and livestock, especially swine. Due to these health concerns, barley used for malting, food or feed is routinely assayed for DON levels. Various methods are available for assaying DON levels in grain samples including enzyme-linked immunosorbent assay (ELISA) and gas chromatography-mass spectrometry (GC-MS). ELISA and GC-MS are very accurate; however, assaying grain samples by these techniques are laborious, expensive and destructive. In this study, we explored the feasibility of using hyperspectral imaging (382-1030 nm) to develop a rapid and non-destructive protocol for assaying DON in barley kernels. Samples of 888 and 116 from various genetic lines were selected for calibration and prediction. Full-wavelength locally weighted partial least squares regression (LWPLSR) achieved high accuracy with the coefficient of determination in prediction (R²_P) of 0.728 and root mean square error of prediction (RMSEP) of 3.802. Competitive adaptive reweighted sampling (CARS) was used to choose potential feature wavelengths, and these selected variables were further optimized using the iterative selection of successive projections algorithm (ISSPA). The CARS-ISSPA-LWPLSR model developed using 7 feature variables yielded R²_P of 0.680 and RMSEP of 4.213 in DON content prediction. Based on the 7 wavelengths selected by CARS-ISSPA, partial least square discriminant analysis (PLSDA) discriminated barley kernels having lower DON (less than1.25 mg/kg) levels from those with higher levels (including 1.25-3 mg/kg, 3-5 mg/kg, and 5-10 mg/kg), with Matthews correlation coefficient in cross-validation (M-R_CV) of as high as 0.931. The results demonstrate that hyperspectral imaging have potential for accelerating non-destructive DON assays of barley samples.

Evaluating the Performance of Newly Integrated Model in Nonlinear Chemical Process Against Missing Measurements

Application of data-driven soft sensors in manufacturing fields, for instance, chemical, pharmaceutical, and bioprocess have rapidly grown. The issue of missing measurements is common in chemical processing industries that involve data-driven soft sensors. Locally weighted Kernel partial least squares (LW-KPLS) algorithm has recently been proposed to develop adaptive soft sensors for nonlinear processes. This algorithm generally works well for complete datasets; however, it is unable to cope well with any datasets comprising missing measurements. Despite the above issue, limited studies can be found in assessing the effects of incomplete data and their treatment method on the predictive performances of LW-KPLS. To address these research gaps, therefore, a trimmed scores regression (TSR) based missing data imputation method was integrated to LW-KPLS to formulate trimmed scores regression assisted locally weighted Kernel partial least squares (TSR-LW-KPLS) model. In this study, this proposed TSR-LW-KPLS was employed to deal with missing measurements in nonlinear chemical process data. The performances of TSR-LW-KPLS were evaluated using three case studies having different percentages of missing measurements varying from 5 % to 40 %. The obtained results were then compared to the results from singular value decomposition assisted locally weighted Kernel partial least squares (SVD-LW-KPLS) model. SVD-LW-KPLS was also proposed by incorporating a singular value decomposition (SVD) based missing data treatment method into LW-KPLS. From the comparative studies, it is evident that the predictive accuracies of TSR-LW-KPLS are superior compared to the ones from SVD-LW-KPLS.

Cleaning verification: A five parameter study of a Total Organic Carbon method development and validation for the cleaning assessment of residual detergents in manufacturing equipment

A Total Organic Carbon (TOC) based analytical method to quantitate trace residues of clean-in-place (CIP) detergents CIP100^® and CIP200^® on the surfaces of pharmaceutical manufacturing equipment was developed and validated. Five factors affecting the development and validation of the method were identified: diluent composition, diluent volume, extraction method, location for TOC sample preparation, and oxidant flow rate. Key experimental parameters were optimized to minimize contamination and to improve the sensitivity, recovery, and reliability of the method. The optimized concentration of the phosphoric acid in the swabbing solution was 0.05M, and the optimal volume of the sample solution was 30mL. The swab extraction method was 1min sonication. The use of a clean room, as compared to an isolated lab environment, was not required for method validation. The method was demonstrated to be linear with a correlation coefficient (R) of 0.9999. The average recoveries from stainless steel surfaces at multiple spike levels were >90%. The repeatability and intermediate precision results were ≤5% across the 2.2-6.6ppm range (50-150% of the target maximum carry over, MACO, limit). The method was also shown to be sensitive with a detection limit (DL) of 38ppb and a quantitation limit (QL) of 114ppb. The method validation demonstrated that the developed method is suitable for its intended use. The methodology developed in this study is generally applicable to the cleaning verification of any organic detergents used for the cleaning of pharmaceutical manufacturing equipment made of electropolished stainless steel material.

Development of Adaptive Soft Sensor Using Locally Weighted Kernel Partial Least Square Model

Locally weighted partial least square (LW-PLS) model has been commonly used to develop adaptive soft sensors and process monitoring for numerous industries which include pharmaceutical, petrochemical, semiconductor, wastewater treatment system and biochemical. The advantages of LW-PLS model are its ability to deal with a large number of input variables, collinearity among the variables and outliers. Nevertheless, since most industrial processes are highly nonlinear, a traditional LW-PLS which is based on a linear model becomes incapable of handling nonlinear processes. Hence, an improved LW-PLS model is required to enhance the adaptive soft sensors in dealing with data nonlinearity. In this work, Kernel function which has nonlinear features was incorporated into LW-PLS model and this proposed model is named locally weighted kernel partial least square (LW-KPLS). Comparisons between LW-PLS and LW-KPLS models in terms of predictive performance and their computational loads were carried out by evaluating both models using data generated from a simulated plant. From the results, it is apparent that in terms of predictive performance LW-KPLS is superior compared to LW-PLS. However, it is found that computational load of LW-KPLS is higher than LW-PLS. After adapting ensemble method with LW-KPLS, computational loads of both models were found to be comparable. These indicate that LW-KPLS performs better than LW-PLS in nonlinear process applications. In addition, evaluation on localization parameter in both LW-PLS and LW-KPLS is also carried out.

A New Local Modelling Approach Based on Predicted Errors for Near-Infrared Spectral Analysis

Over the last decade, near-infrared spectroscopy, together with the use of chemometrics models, has been widely employed as an analytical tool in several industries. However, most chemical processes or analytes are multivariate and nonlinear in nature. To solve this problem, local errors regression method is presented in order to build an accurate calibration model in this paper, where a calibration subset is selected by a new similarity criterion which takes the full information of spectra, chemical property, and predicted errors. After the selection of calibration subset, the partial least squares regression is applied to build calibration model. The performance of the proposed method is demonstrated through a near-infrared spectroscopy dataset of pharmaceutical tablets. Compared with other local strategies with different similarity criterions, it has been shown that the proposed local errors regression can result in a significant improvement in terms of both prediction ability and calculation speed.

Quantitative multiplexing with nano-self-assemblies in SERS

Multiplexed or simultaneous detection of multiple analytes is a valuable tool in many analytical applications. However, complications caused by the presence of interfering compounds in a sample form a major drawback in existing molecular sensor technologies, particularly in multi-analyte systems. Although separating analytes through extraction or chromatography can partially address the problem of interferents, there remains a need for developing direct observational tools capable of multiplexing that can be applied in situ. Surface-enhanced Raman Spectroscopy (SERS) is an optical molecular finger-printing technique that has the ability to resolve analytes from within mixtures. SERS has attracted much attention for its potential in multiplexed sensing but it has been limited in its quantitative abilities. Here, we report a facile supramolecular SERS-based method for quantitative multiplex analysis of small organic molecules in aqueous environments such as human urine.

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.