Near infrared and Raman spectroscopy for the in-process monitoring of pharmaceutical production processes

Luminescence properties and energy transfer of broadband NIR phosphor Li2MgZrO4: 1.0%Cr3+, y%Yb3

The discovery of high thermal stability, broad-band near-infrared (NIR) fluorescent phosphors holds significant potential in applications such as non-destructive testing, promoting plant growth, and night vision devices. In this study, a novel broad-band NIR phosphors Li2MgZrO4 (LMZ): 1.0 %Cr3+, y%Yb3+ were synthesized via a high-temperature solid-state reaction method, with the optimal doping concentration found to be y = 1.5. These phosphors exhibited broad NIR emission in the range of 700-1050 nm by effective energy transfer from Cr3+ to Yb3+. The maximum full width at half maximum (FWHM) of the Cr3+/Yb3+ co-doped LMZ phosphor is 270 nm. The thermal stability of the phosphors was improved with Yb3+ co-doping. Additionally, energy transfer from Cr3+ to Yb3+ was confirmed through luminescence spectra and lifetime analysis. Finally, NIR pc-LED devices composed of a 460 nm ultraviolet chip and LMZ: 1.0 %Cr3+, 1.5 %Yb3+ phosphors were fabricated, offering a highly promising source of invisible light. These results demonstrate the wide-ranging potential applications of this novel, high thermal stability, and ultra-broad NIR emitting fluorescent phosphor.

Is it advantageous to use quality by design (QbD) to develop nanoparticle-based dosage forms for parenteral drug administration?

Parenteral administration is one of the most commonly used drug delivery routes for nanoparticle-based dosage forms, such as lipid-based and polymeric nanoparticles. For the treatment of various diseases, parenteral administration include intravenous, subcutaneous, and intramuscular route. In drug development phase, multiparameter strategy with a focus on drug physicochemical properties and the specificity of the administration route is required. Nanoparticle properties in terms of size and targeted delivery, among others, are able to surpass many drawbacks of conventional dosage forms, but these unique properties can be a bottleneck for approval by regulatory authorities. Quality by Design (QbD) approach has been widely utilized in development of parenteral nanoparticle-based dosage forms. It fosters knowledge of product and process quality by involving sound scientific data and risk assessment strategies. A full and comprehensive investigation into the state of implementation and applications of the QbD approach in these complex drug products can highlight the gaps and challenges. In this review, the analysis of critical attributes and Design of Experiment (DoE) approach in different nanoparticulate systems, together with the proper utilization of Process Analytical Technology (PAT) applications are described. The essential of QbD approach for the design and development of nanoparticle-based dosage forms for delivery via parenteral routes is discussed thoroughly.

Application of electron paramagnetic resonance spectroscopy to examine free radicals formed in indapamide and torasemide storage under UV irradiation and at the higher temperatures which appear under light exposition

Free radical formation in two diuretics: indapamide and torasemide was examined during UV irradiation and storage at higher temperatures using X-band (9.3 GHz) electron paramagnetic resonance spectroscopy (EPR). The aim of this study was to investigate the possibility of storing indapamide and torasemide under UV irradiation and at higher temperatures, which may occur during exposure to light. The diuretic samples were exposed to UVA irradiation for 15, 30 and 45 minutes, and stored at temperatures of 40 °C and 50 °C by 30 minutes. The EPR spectra were analyzed to determine the amplitudes (A), linewidths (ΔBpp), and integral intensities (I) and g factors. The concentrations of free radical (N) in the diuretic samples were also determined. The influence of microwave power on amplitudes, linewidths and the asymmetry parameter were evaluated. The result showed that the tested indapamide and torasemide samples exhibited high free radical concentrations in the range of 1018-1019 spin/g after UV irradiation and heat treatment. Therefore, due to the significant free radical formation indapamide and torasemide should not be stored under UV light and at temperatures of 40 °C and 50 °C. The complex character of free radical systems in the diuretic samples was proved as evidenced by the changes of the asymmetry parameters of the EPR lines with increasing microwave power. Fast spin-lattice relaxation processes were observed in all tested diuretic samples, regardless of the storage conditions. Electron paramagnetic resonance spectroscopy is proposed as a useful method in pharmacy to determine the appropriate storage conditions for diuretics.

Quantification of Hydrogen Isotopes Utilizing Raman Spectroscopy Paired with Chemometric Analysis for Application across Multiple Systems

Online and real-time analysis of a chemical process is a major analytical challenge that can drastically change the way the chemical industry or chemical research operates. With in situ analyses, a new and powerful understanding of chemistry can be gained; however, building robust tools for long-term monitoring faces many challenges, including compensating for instrument drift, instrument replacement, and sensor or probe replacement. Accounting for these changes by recollecting calibration data and rebuilding quantification models can be costly and time-consuming. Here, methods to overcome these challenges are demonstrated with an application of Raman spectroscopy to monitoring hydrogen isotopes with varied speciation within dynamic gas streams. Specifically, chemical data science tools such as chemometric modeling are leveraged along with several examples of calibration transfer approaches. Furthermore, the optimization of instrument and sensor cell parameters for targeted gas-phase analyses is discussed. While the particular focus on hydrogen is highly beneficial within the nuclear energy sector, mechanisms built and demonstrated here are widely applicable to optical spectroscopy monitoring in numerous other chemical systems that can be leveraged in other processes.

Real-Time Quantitative Evaluation of a Drug during Liposome Preparation Using a Probe-Type Raman Spectrometer

During the manufacturing process of liposome formulations, it is considered difficult to evaluate their physicochemical properties and biological profiles due to the complexity of their structure and manufacturing process. Conventional quality evaluation is labor-intensive and time-consuming; therefore, there was a need to introduce a method that could perform in-line, real-time evaluation during the manufacturing process. In this study, Raman spectroscopy was used to monitor in real time the encapsulation of drugs into liposomes and the drug release, which are particularly important quality evaluation items. Furthermore, Raman spectroscopy combined with partial least-squares (PLS) analysis was used for quantitative drug evaluation to assess consistency with results from UV-visible spectrophotometry (UV), a common quantification method. The prepared various ciprofloxacin (CPFX) liposomes were placed in cellulose tubes, and a probe-type Raman spectrophotometer was used to monitor drug encapsulation, the removal of unencapsulated drug, and drug release characteristics in real time using a dialysis method. In the Raman spectra of the liposomes prepared by remote loading, the intensities of the CPFX-derived peaks increased upon drug encapsulation and showed a slight decrease upon removal of the unencapsulated drug. Furthermore, the peak intensity decreased more gradually during the drug release. In all Raman monitoring experiments, the discrepancy between quantified values of CPFX concentration in liposomes, as measured by Raman spectroscopy combined with partial least-squares (PLS) analysis, and those obtained through ultraviolet (UV) spectrophotometry was within 6.7%. The results revealed that the quantitative evaluation of drugs using a combination of Raman spectroscopy and PLS analysis was as accurate as the evaluation using UV spectrophotometry, which was used for comparison. These results indicate the promising potential of Raman spectroscopy as an innovative method for the quality evaluation of liposomal formulations.

Developing a Modular Continuous Drug Product Manufacturing System with Real Time Quality Assurance for Producing Pharmaceutical Mini-Tablets

The pharmaceutical industry has shown keen interest in developing small-scale modular manufacturing systems for producing medicinal products. These systems offer agile and flexible manufacturing, and are well-suited for use in situations requiring rapid production of drugs such as pandemics and humanitarian disasters. The creation of such systems requires the development of modular facilities for making solid oral drug products. In recent years, however, the development of such facilities has seen limited progress. This study presents a development of a prototype modular system that uses drop on demand (DoD) printing to produce personalized solid oral drug products. The system's operation is demonstrated for manufacturing mini-tablets, a category of pediatric drug products, in continuous and semi-batch modes. In this process, the DoD printer is used to generate molten formulation drops that are solidified into mini-tablets. These dosages are then extracted, washed and dried in a continuous filtration and drying unit which is integrated with the printer. Process monitoring tools are also incorporated in the system to track the critical quality attributes of the product and the critical process parameters of the manufacturing operation in real time. Future areas of innovation are also proposed to improve this prototype unit and to enable the development of advanced drug manufacturing systems based on this platform.

Multimodal fiber probe for simultaneous mid-infrared and Raman spectroscopy

A fiber probe has been developed that enables simultaneous acquisition of mid-infrared (MIR) and Raman spectra in the region of 3100-2600 cm-1. Multimodal measurement is based on a proposed ZrO2 crystal design at the tip of an attenuated total reflection (ATR) probe. Mid-infrared ATR spectra are obtained through a pair of chalcogenide infrared (CIR) fibers mounted at the base of the crystal. The probe enables both excitation and acquisition of a weak Raman signal from a portion of the sample in front of the crystal using an additional pair of silica fibers located in a plane perpendicular to the CIR fibers. The advantages of combining MIR and Raman spectra in a single probe have been discussed.

Drug content determination of low-dosed hot-melt extruded filaments using Raman spectroscopy

The aim of this study was to evaluate the suitability of a non-disruptive Raman spectroscopic method to quantify drug concentrations below 5 w% within a polymer matrix produced by hot-melt extrusion (HME). For calibration, praziquantel (PZQ)-polyvinylpyrrolidone-vinylacetat-copolymer (PVP-VA) mixtures were extruded. By focusing the laser light of the Raman probe to a diameter of 1 mm and implementing a self-constructed filament holder, the signal-to-noise (S/N) ratio could be reduced considerably. The obtained Raman spectra show quite high fluorescence, which is likely to be caused by dissolved pharmaceutical active ingredient (API) in the polymer matrix. For content determination, HPLC analysis was conducted as a reference method using the same filament segments. A partial least squares (PLS) model, regressing the PZQ concentrations from HPLC method analysis versus the off-line collected Raman spectra, was developed. The linear correlation for a suitable extrusion run for the production of low-dosed filaments (extrusion 1, two kneading zones) is acceptable (R2 = 0.9915) while the correlation for a extrusion set-up with low miscibility (extrusion 2; without kneading zone) is unacceptable (R2 = 0.5349). The predictive performance of the calibration model from extrusion 1 is rated by the root mean square error of estimation (RMSEE), which was 0.08%. This calibration can now be used to validate the content of low-dosed filaments during HME.

Solid-state analysis for pharmaceuticals: Pathways to feasible and meaningful analysis

The solid state of matter is the preferred starting point for designing a pharmaceutical product. This is driven by both patient preferences and the relative ease of supplying a solid pharmaceutical product with desired quality and performance. Solid form diversity is increasingly prevalent as a crucial element in designing these products, which underpins the importance of solid-state analytical methods. This paper provides a critical analysis of challenges related to solid-state analytics, as well as considerations and suggestions for feasible and meaningful pharmaceutical analysis.

UV/Vis spectroscopy as an in-line monitoring tool for tablet content uniformity

Continuous manufacturing provides advantages compared to batch manufacturing and is increasingly gaining importance in the pharmaceutical industry. In particular, the implementation of tablet processes in continuous plants is an important part of current research. For this, in-line real-time monitoring of product quality through process analytical technology (PAT) tools is crucial. This study focuses on an in-line UV/Vis spectroscopy method for monitoring the active pharmaceutical ingredient (API) content in tablets. UV/Vis spectroscopy is particularly advantageous here, because it allows univariate data analysis without complex data processing. Experiments were conducted on a rotary tablet press. The tablets consisted of 7- 13 wt% theophylline monohydrate as API, lactose monohydrate and magnesium stearate. Two tablet production rates were investigated, 7200 and 20000 tablets per hour. The UV/Vis probe was mounted at the ejection position and measurements were taken on the tablet sidewall. Validation was according to ICH Q2 with respect to specificity, linearity, precision, accuracy and range. The specificity for this formulation was proven and linearity was sufficient with coefficients of determination of 0.9891 for the low throughput and 0.9936 for the high throughput. Repeatability and intermediate precision were investigated. Both were sufficient, indicated by coefficients of variations with a maximum of 6.46% and 6.34%, respectively. The accuracy was evaluated by mean percent recovery. This showed a higher accuracy at 20000 tablets per hour than 7200 tablets per hour. However, both throughputs demonstrate sufficient accuracy. Finally, UV/Vis spectroscopy is a promising alternative to the common NIR and Raman Spectroscopy.

Characterization of enteric-coated erythromycin tablets by Raman mapping and its pharmaceutical evaluation

Enteric tablet coating thickness is a critical quality attribute of the coating process that can affect dissolution behavior in vitro as well as release in vivo. Raman mapping offers unique advantages in analyzing the distribution of active pharmaceutical ingredients and excipients in formulations. In this study, Raman mapping was used to characterize the coating of enteric-coated erythromycin tablets coated by two different processes and compare the differences in their coating formulation, thickness, and uniformity. Furthermore, we aimed to select the appropriate pH of the dissolution medium at which the coating slowly cracks to release the drug and determine the dissolution profile. The differences in the coating thickness and uniformity of the two products resulted in differences in dissolution behavior. Although there are differences in the coating processes for the two types of enteric-coated erythromycin tablets, the thickness of the outer coating on the side is a critical quality attribute in both processes. The outer coating of product A is relatively thick, and the thickness of the outer coating on the side affects the dissolution amount. The outer coating of product B is relatively thin, resulting in a short cracking time and large variation and a significant difference in the initial dissolution amounts between tablets. Raman mapping can be used to analyze the differences in coating formulations and for process evaluation.

Highly efficient near-infrared solid solution phosphors with excellent thermal stability and tunable spectra for pc-LED light sources toward NIR spectroscopy applications

Near-infrared (NIR) luminescent materials have attracted wide research interest due to their unique photophysical properties for designing NIR light-emitting diodes (NIR LEDs). Here, a series of Cr3+-activated NIR-emitting solid solution phosphors, Gd1-xLux(Al1-xScx)3(BO3)4:0.01Cr3+ (GLASB:Cr3+) (x = 0 to 0.5), are successfully synthesized via a cosubstitution approach. The GLASB:Cr3+ phosphors reveal extraordinary optical performance with a desirable high IQE of 93.6%, considerable broadened FWHM (from 128 nm to 196 nm) and redshift of 119 nm (747 → 866 nm) as the amount of [Lu3+-Sc3+] ion doping increases. Moreover, their photoluminescent thermal stability is substantially improved, maintaining 105.7% of the initial integral intensity up to 150 °C, namely zero-thermal-quenching. The NIR pc-LED fabricated using the GLASB:Cr3+ phosphor generates an NIR output power of 46 mW and an electro-optical efficiency of 37% at a 120 mA input current. Finally, the characteristic NIR emission of this phosphor can not only be utilized in the fields of night-vision technology and biometric identification, but also exhibits a perfect match with the absorption of the bacteriochlorophyll (BChl) and light-harvesting protein (LHP) of photosynthetic bacteria (PSB), presenting a high application prospect for improving PSB photosynthesis.

Label-Free Quantification of Nanoencapsulated Piperonyl Esters in Cosmetic Hydrogels Using Raman Spectroscopy

Raman spectroscopy is a well-established technique for the molecular characterisation of samples and does not require extensive pre-analytical processing for complex cosmetic products. As an illustration of its potential, this study investigates the quantitative performance of Raman spectroscopy coupled with partial least squares regression (PLSR) for the analysis of Alginate nanoencapsulated Piperonyl Esters (ANC-PE) incorporated into a hydrogel. A total of 96 ANC-PE samples covering a 0.4% w/w-8.3% w/w PE concentration range have been prepared and analysed. Despite the complex formulation of the sample, the spectral features of the PE can be detected and used to quantify the concentrations. Using a leave-K-out cross-validation approach, samples were divided into a training set (n = 64) and a test set, samples that were previously unknown to the PLSR model (n = 32). The root mean square error of cross-validation (RMSECV) and prediction (RMSEP) was evaluated to be 0.142% (w/w PE) and 0.148% (w/w PE), respectively. The accuracy of the prediction model was further evaluated by the percent relative error calculated from the predicted concentration compared to the true value, yielding values of 3.58% for the training set and 3.67% for the test set. The outcome of the analysis demonstrated the analytical power of Raman to obtain label-free, non-destructive quantification of the active cosmetic ingredient, presently PE, in complex formulations, holding promise for future analytical quality control (AQC) applications in the cosmetics industry with rapid and consumable-free analysis.

Towards Point-of-Care Manufacturing and Analysis of Immediate-Release 3D Printed Hydrocortisone Tablets for The Treatment of Congenital Adrenal Hyperplasia

Hydrocortisone (HC) is the preferred drug in children with congenital adrenal hyperplasia due to its lower potency as well as fewer reports of side effects. Fused deposition modelling (FDM) 3D printing holds the potential to produce low-cost personalised doses for children at the point of care. However, the compatibility of the thermal process to produce immediate-release bespoke tablets for this thermally labile active is yet to be established. This work aims to develop immediate-release HC tablets using FDM 3D printing and assess drug contents as a critical quality attribute (CQA) using a compact, low-cost near-infrared (NIR) spectroscopy as a process analytical technology (PAT). The FDM 3D printing temperature (140 °C) and drug concentration in the filament (10%-15% w/w) were critical parameters to meet the compendial criteria for drug contents and impurities. Using a compact low-cost NIR spectral device over a wavelength of 900-1700 nm, the drug contents of 3D printed tablets were assessed. Partial least squares (PLS) regression was used to develop individual calibration models to detect HC content in 3D printed tablets of lower drug contents, small caplet design, and relatively complex formula. The models demonstrated the ability to predict HC concentrations over a wide concentration range (0-15% w/w), which was confirmed by HPLC as a reference method. Ultimately, the capability of the NIR model had preceding dose verification performance on HC tablets, with linearity (R² = 0.981) and accuracy (RMSECV = 0.46%). In the future, the integration of 3DP technology with non-destructive PAT techniques will accelerate the adoption of on-demand, individualised dosing in a clinical setting.

The application of Near-Infrared Spatially Resolved Spectroscopy in scope of achieving continuous real-time quality monitoring and control of tablets with challenging dimensions

In scope of achieving real-time release of tablets, quality attributes need to be monitored and controlled through Process Analytical Technology tools such as near-infrared spectroscopy (NIRS). The authors evaluated the suitability of NIR-Spatially Resolved Spectroscopy (NIR-SRS) for continuous real-time monitoring and control of content uniformity, hardness and homogeneity of tablets with challenging dimensions. A novel user-friendly research and development inspection unit was used as standalone equipment for the analysis of small oblong tablets with deep-cut break lines. A total of 66 tablets varying in hardness and Active Pharmaceutical Ingredient (API) content were inspected, with each tablet being analysed five times and measurements repeated on three different days. Partial Least Squares (PLS) models were developed to assess content uniformity and hardness, of which the former showed higher accuracy. The authors attempted to visualize tablet homogeneity through NIR-SRS spectra by regressing all spectra obtained during a single measurement using a content uniformity PLS model. The NIR-SRS probe demonstrated its potential towards real-time release testing through its ability to quickly monitor content uniformity, hardness and visualize homogeneity, even for tablets with challenging dimensions.

Use of machine learning tools and NIR spectra to estimate residual moisture in freeze-dried products

Residual Moisture (RM) in freeze-dried products is one of the most important critical quality attributes (CQAs) to monitor, since it affects the stability of the active pharmaceutical ingredient (API). The standard experimental method adopted for the measurements of RM is the Karl-Fischer (KF) titration, that is a destructive and time-consuming technique. Therefore, Near-Infrared (NIR) spectroscopy was widely investigated in the last decades as an alternative tool to quantify the RM. In the present paper, a novel method was developed based on NIR spectroscopy combined with machine learning tools for the prediction of RM in freeze-dried products. Two different types of models were used: a linear regression model and a neural network based one. The architecture of the neural network was chosen so as to optimize the prediction of the residual moisture, by minimizing the root mean square error with the dataset used in the learning step. Moreover, the parity plots and the absolute error plots were reported, allowing a visual evaluation of the results. Different factors were considered when developing the model, namely the range of wavelengths considered, the shape of the spectra and the type of model. The possibility of developing the model using a smaller dataset, obtained with just one product, that could be then applied to a wider range of products was investigated, as well as the performance of a model developed for a dataset encompassing several products. Different formulations were analyzed: the main part of the dataset was characterized by a different percentage of sucrose in solution (3%, 6% and 9% specifically); a smaller part was made up of sucrose-arginine mixtures at different percentages and only one formulation was characterized by another excipient, the trehalose. The product-specific model for the 6% sucrose mixture was found consistent for the prediction of RM in other sucrose containing mixtures and in the one containing trehalose, while failed for the dataset with higher percentage of arginine. Therefore, a global model was developed by including a certain percentage of all the available dataset in the calibration phase. Results presented and discussed in this paper demonstrate the higher accuracy and robustness of the machine learning based model with respect to the linear models.

Assessment of blend uniformity in a stream sampler device using Raman spectroscopy

This study describes the first implementation of Raman spectrometer in a stream sampler for the in-line monitoring of low drug concentration in poor flowability powder blends. Raman spectra were continuously acquired as the powder blends flowed through the stream sampler operating with a paddle wheel speed of 10 RPM and used to develop the calibration models. A calibration model was developed to quantify caffeine concentration from 1.50 to 4.50% w/w using Partial Least Squares Regression (PLS-R). Three test set blends were used to assess the prediction errors of the calibration model. Caffeine concentration was predicted for the test set blends with a root mean square error of prediction of 0.21% w/w and a low bias of -0.03% w/w. The calibration model showed good prediction performance with an estimated sample mass of 83 mg. Variographic analysis demonstrated the low process variance of the real-time spectral acquisition through minimum practical error and sill values. The results showed the ability of the Raman spectrometer coupled with the stream sampler to monitor low drug concentration for poor flowability blends.

Towards a real-time release of blends and tablets using NIR and Raman spectroscopy at commercial scales

Near Infrared and Raman spectroscopy-based Process Analytical Technology tools were used for monitoring blend uniformity (BU) and content uniformity (CU) for solid oral formulations. A quantitative Partial Least Square model was developed to monitor BU as real-time release testing at a commercial scale. The model having the R2, and root mean square error of 0.9724 and 2.2047, respectively can predict the target concentration of 100% with a 95% confidence interval of 101.85-102.68% even after one year. The tablets from the same blends were investigated for CU using NIR and Raman techniques both in reflection and transmission mode. Raman reflection technique was found to be the best and the PLS model was developed using tablets compressed at different concentrations, hardness, and speed. The model with R2 and RMSE of 0.9766 and 1.9259, respectively was used for the quantification of CU. Both the BU and CU models were validated for accuracy, precision, specificity, linearity, and robustness. The accuracy was proved against the HPLC method with a relative standard deviation of less than 3%. The equivalency for BU by NIR and CU by Raman was evaluated using Schuirmann's Two One-sided tests and found equivalent to HPLC within a 2% acceptable limit.

Interpretable artificial neural networks for retrospective QbD of pharmaceutical tablet manufacturing based on a pilot-scale developmental dataset

As the pharmaceutical industry increasingly adopts the Pharma 4.0. concept, there is a growing need to effectively predict the product quality based on manufacturing or in-process data. Although artificial neural networks (ANNs) have emerged as powerful tools in data-rich environments, their implementation in pharmaceutical manufacturing is hindered by their black-box nature. In this work, ANNs were developed and interpreted to demonstrate their applicability to increase process understanding by retrospective analysis of developmental or manufacturing data. The in vitro dissolution and hardness of extended-release, directly compressed tablets were predicted from manufacturing and spectroscopic data of pilot-scale development. The ANNs using material attributes and operational parameters provided better results than using NIR or Raman spectra as predictors. ANNs were interpreted by sensitivity analysis, helping to identify the root cause of the batch-to-batch variability, e.g., the variability in particle size, grade, or substitution of the hydroxypropyl methylcellulose excipient. An ANN-based control strategy was also successfully utilized to mitigate the batch-to-batch variability by flexibly operating the tableting process. The presented methodology can be adapted to arbitrary data-rich manufacturing steps from active substance synthesis to formulation to predict the quality from manufacturing or development data and gain process understanding and consistent product quality.

Developing a Virtual Flowability Sensor for Monitoring a Pharmaceutical Dry Granulation Line

Current technologies to measure granule flowability involve at-line methods that can take hours to perform. This is problematic for a continuous dry granulation tableting line, where the quality assurance and control of the final tablet products depend on real-time monitoring and control of powder flowability. Hence, a real-time alternative is needed for measuring the flowability of the granular products coming out of the roller compactor, which is the unit operation immediately preceding the tablet press. Since particle analyzers have the potential to take inline measurements of the size and shape of granules, they can potentially serve as real-time flowability sensors, given that the size and shape measurements can be used to reliably predict flowability measurements. This paper reports on the use of Partial Least Squares (PLS) regression to utilize distributions of size and shape measurements in predicting the output of three different types of flowability measurements: rotary drum flow, orifice flow, and tapped density analysis. The prediction performance of PLS had a coefficient of determination ranging from 0.80 to 0.97, which is the best reported performance in the literature. This is attributed to the ability of PLS to handle high collinearity in the datasets and the inclusion of multiple shape characteristics-eccentricity, form factor, and elliptical form factor-into the model. The latter calls for a change in industry perspective, which normally dismisses the importance of shape in favor of size; and the former suggests the use of PLS as a better way to reduce the dimensionality of distribution datasets, instead of the widely used practice of pre-selecting distribution percentiles.

Mannitol as an Excipient for Lyophilized Injectable Formulations

The review summarizes the current state of knowledge of mannitol as an excipient in lyophilized injectable small and large molecule formulations. When compared with glycine, the physicochemical properties of mannitol make it a desirable and preferred bulking agent. Though mannitol is a popular bulking agent in freeze-dried formulations, its use may pose certain challenges such as vial breakage or its existence as a metastable crystalline hemihydrate in the final cake, necessitating appropriate mitigation strategies. The understanding of the phase behavior of mannitol in aqueous systems, during the various stages of freeze-drying, can be critical for the optimization of freeze-drying cycle parameters in multi-component formulations. Finally, using a decision tree as a guiding tool, we demonstrate the use of orthogonal techniques for attaining a stable and cost-effective lyophilized drug product containing mannitol.

Liposome manufacturing under continuous flow conditions: towards a fully integrated set-up with in-line control of critical quality attributes

Continuous flow manufacturing (CFM) has shown remarkable advantages in the industrial-scale production of drug-loaded nanomedicines, including mRNA-based COVID-19 vaccines. Thus far, CFM research in nanomedicine has mainly focused on the initial particle formation step, while post-formation production steps are hardly ever integrated. The opportunity to implement in-line quality control of critical quality attributes merits closer investigation. Here, we designed and tested a CFM setup for the manufacturing of liposomal nanomedicines that can potentially encompass all manufacturing steps in an end-to-end system. Our main aim was to elucidate the key composition and process parameters that affect the physicochemical characteristics of the liposomes. Total flow rate, lipid concentration and residence time of the liposomes in a high ethanol environment (i.e., above 20% v/v) emerged as critical parameters to tailor liposome size between 80 and 150 nm. After liposome formation, the pressure and the surface area of the filter in the ultrafiltration unit were critical parameters in the process of clearing the dispersion from residual ethanol. As a final step, we integrated in-line measurement of liposome size and residual ethanol content. Such in-line measurements allow for real-time monitoring and in-process adjustment of key composition and process parameters.

Emerging PAT for Freeze-Drying Processes for Advanced Process Control

Lyophilization is a widely used drying operation, but long processing times are a major drawback. Most lyophilization processes are conducted by a recipe that is not changed or optimized after implementation. With the regulatory demanded quality by design (QbD) approach, the process can be controlled inside an optimal range, ensuring safe process conditions. Process analytical technology (PAT) is crucial because it allows real-time monitoring and is part of a control strategy. In this work, emerging PAT (manometric temperature measurement (MTM), comparative pressure measurement, heat flux sensors, and ice ruler) are used for measurements during the freeze-drying process, and their potential for implementation inside a control strategy is outlined.

Soft sensor for content prediction in an integrated continuous pharmaceutical formulation line based on the residence time distribution of unit operations

In this study, a concentration predicting soft sensor was achieved based on the Residence Time Distribution (RTD) of an integrated, three-step pharmaceutical formulation line. The RTD was investigated with color-based tracer experiments using image analysis. Twin-screw wet granulation (TSWG) was directly coupled with a horizontal fluid bed dryer and an oscillating mill. Based on integrated measurement, we proved that it is also possible to couple the unit operations in silico. Three surrogate tracers were produced with a coloring agent to investigate the separated unit operations and the solid and liquid inputs of the TSWG. The soft sensor's prediction was compared to validating experiments of a 0.05 mg/g (15% of the nominal) concentration change with High-Performance Liquid Chromatography (HPLC) reference measurements of the active ingredient proving the adequacy of the soft sensor (RMSE < 4%).

Development of Inline Near-Infrared Spectroscopy Method for Real-Time Monitoring of Blend Uniformity of Direct Compression and Granulation-Based Products at Commercial Scales

Blending is a critical intermediate unit operation for all solid oral formulations. For blend uniformity testing, API content in the blend must be quantified precisely. A detailed study was conducted to demonstrate the suitability of inline NIR (near-infrared) spectroscopy for blend uniformity testing of two solid oral formulations: existing direct compression (DC) product with a multistep blending process and granulation-based product with API granules. Both qualitative and quantitative methods were developed at a laboratory scale using statistical moving block standard deviation (MBSD) and multivariate data analysis such as principal component analysis (PCA) and partial least squares (PLS) regression. The qualitative MBSD method demonstrated that there was no need for multiple steps for the existing DC product. Hence, a simplified single-step process was developed for blending. Quantitative PLS models for blending processes of both the products were developed, validated, and successfully implemented at a commercial scale for the real-time release of blends. Results obtained from the validated model were in good agreement with the current method of sampling and chromatography.

A Review of Pharmaceutical Robot based on Hyperspectral Technology

The quality and safety of medicinal products are related to patients’ lives and health. Therefore, quality inspection takes a key role in the pharmaceutical industry. Most of the previous solutions are based on machine vision, however, their performance is limited by the RGB sensor. The pharmaceutical visual inspection robot combined with hyperspectral imaging technology is becoming a new trend in the high-end medical quality inspection process since the hyperspectral data can provide spectral information with spatial knowledge. Yet, there is no comprehensive review about hyperspectral imaging-based medicinal products inspection. This paper focuses on the pivotal pharmaceutical applications, including counterfeit drugs detection, active component analysis of tables, and quality testing of herbal medicines and other medical materials. We discuss the technology and hardware of Raman spectroscopy and hyperspectral imaging, firstly. Furthermore, we review these technologies in pharmaceutical scenarios. Finally, the development tendency and prospect of hyperspectral imaging technology-based robots in the field of pharmaceutical quality inspection is summarized.

A review of ultrasonic sensing and machine learning methods to monitor industrial processes

Supervised machine learning techniques are increasingly being combined with ultrasonic sensor measurements owing to their strong performance. These techniques also offer advantages over calibration procedures of more complex fitting, improved generalisation, reduced development time, ability for continuous retraining, and the correlation of sensor data to important process information. However, their implementation requires expertise to extract and select appropriate features from the sensor measurements as model inputs, select the type of machine learning algorithm to use, and find a suitable set of model hyperparameters. The aim of this article is to facilitate implementation of machine learning techniques in combination with ultrasonic measurements for in-line and on-line monitoring of industrial processes and other similar applications. The article first reviews the use of ultrasonic sensors for monitoring processes, before reviewing the combination of ultrasonic measurements and machine learning. We include literature from other sectors such as structural health monitoring. This review covers feature extraction, feature selection, algorithm choice, hyperparameter selection, data augmentation, domain adaptation, semi-supervised learning and machine learning interpretability. Finally, recommendations for applying machine learning to the reviewed processes are made.

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.

Aerogel-Lined Capillaries for Raman Signal Gain of Aqueous Mixtures

We report an experimental study on the gain of the Raman signal of aqueous mixtures and liquid water when confined in aerogel-lined capillaries of various lengths of up to 20 cm and various internal diameters between 530 and 1000 µm. The lining was made of hydrophobised silica aerogel, and the carrier capillary body consisted of fused silica or borosilicate glass. Compared to the Raman signal detected from bulk liquid water with the same Raman probe, a Raman signal 27 times as large was detected when the liquid water was confined in a 20 cm-long capillary with an internal diameter of 700 µm. In comparison with silver-lined capillaries of the same length and same internal diameter, the aerogel-lined capillaries featured a superior Raman signal gain and a longer gain stability when exposed to mixtures of water, sugar, ethanol and acetic acid.

Method development and validation of a near-infrared spectroscopic method for in-line API quantification during fluidized bed granulation

Near-infrared spectroscopy (NIRS) is an excellent process analytical technology (PAT) tool for active pharmaceutical ingredient (API) quantification during fluidized granulation. Therefore, a portable near-infrared spectrometer combined with a new innovative method of extended iterative optimization technique (EIOT) was used to in-line monitor the API content uniformity during fluidized bed granulation. The principal component analysis (PCA) and partial least squares regression (PLSR) were also used to characterize and predict API concentration with changes from 75% to 125% of the label claim to prove the superiority of EIOT. The API content prediction accuracy of the EIOT method was verified through offline High Performance Liquid Chromatography (HPLC) measurement. Also, the spatial distribution of API in granules was visualized by Raman imaging technology. The results showed that the established NIRS method was suitable for the prediction of API content in fluidized bed granulation, which provides a new idea for the determination of API content during granulation.

Broadband Near-Infrared-Emitting Phosphors with Suppressed Concentration Quenching in a Two-Dimensional Structure

For inorganic luminescent materials with activators, the energy yield is usually observed to decrease with an increase in activator concentration, which is known as the concentration quenching effect. To inhibit this phenomenon, a common strategy is to increase the distance between activators. Most previous reports have focused on the three-dimensional crystal lattice, and there have been few reports about two-dimensional layered structure. Herein, we synthesized a novel Cr³⁺-activated near-infrared (NIR) phosphor Li₂Sr₂Al(PO₄)₃ (LSAPO) with layered structure, and in such a two-dimensional structure, we proved experimentally that the concentration quenching was suppressed. Under 460 nm excitation, LSAPO:Cr³⁺ gave a broad NIR emission band (700-1200 nm) centered at 823 nm with a full width at half-maximum (fwhm) of 178 nm and a broad absorption band, indicating its potential application in NIR spectroscopy. Moreover, by codoping Cr³⁺ and Yb³⁺ ions, we further widened the emission bandwidth to ∼230 nm of fwhm, the internal quantum efficiency increased from 54% to 61%, and the thermal stability was improved. The fabricated NIR device with a LSAPO:Cr³⁺,Yb³⁺ phosphor coupled with blue chips can be applied in night-vision technologies and medical fields.

A review on the modernization of pharmaceutical development and manufacturing - Trends, perspectives, and the role of mathematical modeling

Recently, the pharmaceutical industry has been facing several challenges associated to the use of outdated development and manufacturing technologies. The return on investment on research and development has been shrinking, and, at the same time, an alarming number of shortages and recalls for quality concerns has been registered. The pharmaceutical industry has been responding to these issues through a technological modernization of development and manufacturing, under the support of initiatives and activities such as quality-by-design (QbD), process analytical technology, and pharmaceutical emerging technology. In this review, we analyze this modernization trend, with emphasis on the role that mathematical modeling plays within it. We begin by outlining the main socio-economic trends of the pharmaceutical industry, and by highlighting the life-cycle stages of a pharmaceutical product in which technological modernization can help both achieve consistently high product quality and increase return on investment. Then, we review the historical evolution of the pharmaceutical regulatory framework, and we discuss the current state of implementation and future trends of QbD. The pharmaceutical emerging technology is reviewed afterwards, and a discussion on the evolution of QbD into the more effective quality-by-control (QbC) paradigm is presented. Further, we illustrate how mathematical modeling can support the implementation of QbD and QbC across all stages of the pharmaceutical life-cycle. In this respect, we review academic and industrial applications demonstrating the impact of mathematical modeling on three key activities within pharmaceutical development and manufacturing, namely design space description, process monitoring, and active process control. Finally, we discuss some future research opportunities on the use of mathematical modeling in industrial pharmaceutical environments.

Comparative study on the real-time monitoring of a fluid bed drying process of extruded granules using near-infrared spectroscopy and audible acoustic emission

The process of fluidized bed drying of granules was comparatively evaluated by on-line real-time measurements of granule moisture content (MC) using near-infrared spectroscopy (NIR) and audible acoustic emission (AAE). The extruded granules were prepared by kneading a powder blend containing lactose, starch, crystalline cellulose, and riboflavin, with water. The MC of the granules (while they were dried at 35 °C in a fluidized bed dryer) was monitored simultaneously with NIR and AAE. The prediction accuracy of the NIR and AAE using partial least squares (PLS) was verified by measuring MC of the granules. The best calibration models following NIR and AAE evaluations consisted of five latent variables with correlation coefficients of 1.000 and 0.998 and root mean square error of 0.259 and 0.615, respectively. As a result of external verification, the accuracy of MC analysis by AAE was slightly lower than that of NIR; however, it was still applicable in practice. Furthermore, the end point of fluidized bed drying process was automatically determined using the PLS discriminant analysis. From the above results, it can be concluded that the AAE-mediated granule drying process can be monitored with sufficient accuracy (compared with NIR).

Combined Raman and Turbidity Probe for Real-Time Analysis of Variable Turbidity Streams

Real-time and in situ process monitoring is a powerful tool that can empower operators of hazardous processes to better understand and control their chemical systems without increased risk to themselves. However, the application of monitoring techniques to complex chemical processes can face challenges. An example of this is the application of optical spectroscopy, otherwise capable of providing detailed chemical composition information, to processes exhibiting variable turbidity. Here, details on a novel combined Raman spectroscopy and turbidimetry probe are discussed, which advances current technology to enable flexible and robust in situ monitoring of a flowing process stream. Furthermore, the analytical approach to accurately account for both Raman signal and turbidity while quantifying chemical targets is detailed. This new approach allows for accurate analysis without requiring assumptions of stable process chemistry, which may be unlikely in applications such as waste cleanup. Through leveraging Raman and turbidity data simultaneously collected from the combined probe within chemometric models, accurate quantification of multiple chemical targets can be achieved under conditions of variable concentrations and turbidity.

A unique approach for in-situ monitoring of the THCA decarboxylation reaction in solid state

The decarboxylation of Δ9-tetrahydrocannabinolic acid (THCA) plays pivotal role in the potency of medical cannabis and its extracts. Our present work aims to draw attention to mid-infrared (MIR) spectroscopy to in-situ monitor and decipher the THCA decarboxylation reaction in the solid state. The initial TG/DTG curves of THCA, for a first time, outlined the solid-solid decarboxylation dynamics, defined the endpoint of the process and the temperature of the maximal conversion rate, which aided in the design of the further IR experiment. Temperature controlled IR spectroscopy experiments were performed on both THCA standard and cannabis flower by providing detailed band assignment and conducting spectra-structure correlations, based on the concept of functional groups vibrations. Moreover, a multivariate statistical analysis was employed to address the spectral regions of utmost importance for the THCA → THC interconversion process. The principal component analysis model was reduced to two PCs, where PC1 explained 94.76% and 98.21% of the total spectral variations in the THCA standard and in the plant sample, respectively. The PC1 plot score of the THCA standard, as a function of the temperature, neatly complemented to the TG/DTG curves and enabled determination of rate constants for the decarboxylation reaction undertaken on several selected temperatures. The predictive capability of MIR was further demonstrated with PLS (R2X = 0.99, R2Y = 0.994 and Q2 = 0.992) using thermally treated flower samples that covered broad range of THCA/THC content. Consequently, a progress in elucidation of kinetic models of THCA decarboxylation in terms of fitting the experimental data for both, solid state standard substance and a plant flower, was achieved. The results open the horizon to promote an appropriate process analytical technology (PAT) in the outgrowing medical cannabis industry.

Near-infrared analysis of nanofibrillated cellulose aerogel manufacturing

Biomaterial aerogel fabrication by freeze-drying must be further improved to reduce the costs of lengthy freeze-drying cycles and to avoid the formation of spongy cryogels and collapse of the aerogel structures. Residual water content is a critical quality attribute of the freeze-dried product, which can be monitored in-line with near-infrared (NIR) spectroscopy. Predictive models of NIR have not been previously applied for biomaterials and the models were mostly focused on the prediction of only one formulation at a time. We recorded NIR spectra of different nanofibrillated cellulose (NFC) hydrogel formulations during the secondary drying and set up a partial least square regression model to predict their residual water contents. The model can be generalized to measure residual water of formulations with different NFC concentrations and the excipients, and the NFC fiber concentrations and excipients can be separated with the principal component analysis. Our results provide valuable information about the freeze-drying of biomaterials and aerogel fabrication, and how NIR spectroscopy can be utilized in the optimization of residual water content.

Layer thickness prediction and tissue classification in two-layered tissue structures using diffuse reflectance spectroscopy

During oncological surgery, it can be challenging to identify the tumor and establish adequate resection margins. This study proposes a new two-layer approach in which diffuse reflectance spectroscopy (DRS) is used to predict the top layer thickness and classify the layers in two-layered phantom and animal tissue. Using wavelet-based and peak-based DRS spectral features, the proposed method could predict the top layer thickness with an accuracy of up to 0.35 mm. In addition, the tissue types of the first and second layers were classified with an accuracy of 0.95 and 0.99. Distinguishing multiple tissue layers during spectral analyses results in a better understanding of more complex tissue structures encountered in surgical practice.

Process Analytical Technologies - Advances in bioprocess integration and future perspectives

Process Analytical Technology (PAT) instruments include analyzers capable of measuring physical and chemical process parameters and key attributes with the goal of optimizing process controls. PAT in the form of a probe or sensor is designed to integrate within the pharmaceutical manufacturing line and is coupled with computing equipment to perform chemometric modeling for result interpretation and multilayer statistical control of processes. PAT solutions are intended for understanding bioprocesses with a goal to control quality at all stages of product manufacturing and achieve quality by design (QbD). The goal of PAT implementation is to promote real-time release of products to decrease the cycle time and cost of production. This review focuses on the applications of PAT solutions at different stages of the manufacturing process for vaccine production, the advantages, challenges at present state, and the vision of the future development of biopharmaceutical industries.

Rapid determination of process parameters during simultaneous saccharification and fermentation (SSF) of cassava based on molecular spectral fusion (MSF) features

Simultaneous saccharification and fermentation (SSF) of cassava is one of the key steps in the production of fuel ethanol. In order to improve the monitoring efficiency of the ethanol production process and the product yield, this study puts forward a new idea for monitoring of the cassava SSF process based on the molecular spectroscopy fusion (MSF) technique. Savisky-Golay (SG) combined with standard normal variable (SNV) was used to preprocess the obtained Raman spectra and near-infrared (NIR) spectra. Competitive adaptive reweighted sampling (CARS) was used to optimize the characteristic wavelengths of the preprocessed Raman spectra and the NIR spectra, and the optimized features were fused in the feature layer. The support vector machine (SVM) model of the process parameters during the cassava SSF based on the MSF features was established. The experimental results showed that compared with the best CARS-SVM model based on the single-molecule spectral features, the performance of the best CARS-SVM model based on fusion features has been significantly improved. For detection of the glucose content, the RMSEP, R_P² and RPD of the best CARS-SVM model were 5.398, 0.957 and 4.922, respectively. For detection of the ethanol content, the RMSEP, R_P² and RPD of the best CARS-SVM model were 4.394, 0.977 and 6.758, respectively. The obtained results reveal that the combination of MSF technique and appropriate chemometric methods can achieve high-precision quantitative detection of the process parameters during the cassava SSF. This study can provide technical basis and experimental reference for the development of portable spectrometer equipment for process monitoring of the cassava SSF.

Utilizing in situ spectroscopic tools to monitor ketal deprotection processes

The use of protection groups to shield a functional group during a synthesis is employed throughout many reactions and organic syntheses. The role of a protection group can be vital to the success of a reaction, as well as increase reaction yield and selectivity. Although much work has been done to investigate the addition of a protection group, the removal of the protection group is just as important - however, there is a lack of methods employed within the literature for monitoring the removal of a protection group in real time. In this work, the process of removing, or deprotecting, a ketal protecting group is investigated. Process analytical technology tools are incorporated for in situ analysis of the deprotection reaction of a small molecule model compound. Specifically, Raman spectroscopy and Fourier transform infrared spectroscopy show that characteristic bands can be used to track the decrease of the reactant and the increase of the expected products over time. To the best of our knowledge, this is the first report of process analytical technology being used to monitor a ketal deprotection reaction in real time. This information can be capitalized on in the future for understanding and optimizing pharmaceutically-relevant deprotection processes and downstream reactions.

Cr3+-Doped double perovskite antimonates: efficient and tunable phosphors from NIR-I to NIR-II

By tuning cationic compositions, Cr3+-doped antimonate double perovskites present emissions from NIR-I to NIR-II regions. In particular, Ca2ScSbO6:Cr3+ and Sr2InSbO6:Cr3+ feature long-wavelength emissions with high luminescence efficiency.

Quantitative analysis of blend uniformity within a Three-Chamber feed frame using simultaneously Raman and Near-Infrared spectroscopy

This study reports the use of Raman and Near-infrared (NIR) spectroscopy to simultaneously monitor the drug concentration in flowing powder blends within a three-chamber feed frame. The Raman probe was located at the top of the dosing chamber, while the NIR probe was located at the top of the filling chamber. The Raman and NIR spectra were continuously acquired while the powder blends flowed through the feed frame. Calibration models were developed with spectra from a total of five calibration blends ranging in caffeine concentration among 3.50 and 6.50% w/w. These models were optimized to predict three test set blends of 4.00, 5.00, and 6.00% w/w caffeine. The results showed a high predictive ability of the models based on root mean square error of predictions of 0.174 and 0.235% w/w for NIR and Raman spectroscopic models, respectively. Concentration profiles with higher variability were observed for the Raman spectroscopy predictions. An estimate of the mass analyzed by each spectrum showed that a NIR spectrum analyzes approximately 4.5 times the mass analyzed by a Raman spectrum; despite these differences in the mass analyzed, blend uniformity results are equivalent between techniques. Variographic analysis demonstrated that both techniques have significantly low sampling errors for the real-time monitoring process of drug concentration within the feed frame.

A FT-NIR Process Analytical Technology Approach for Milk Renneting Control

The study proposes a process analytical technology (PAT) approach for the control of milk coagulation through near infrared spectroscopy (NIRS), computing multivariate statistical process control (MSPC) charts, based on principal component analysis (PCA). Reconstituted skimmed milk and commercial pasteurized skimmed milk were mixed at two different ratios (60:40 and 40:60). Each mix ratio was prepared in six replicates and used for coagulation trials, monitored by fundamental rheology, as a reference method, and NIRS by inserting a probe directly in the coagulation vat and collecting spectra at two different acquisition times, i.e., 60 s or 10 s. Furthermore, three failure coagulation trials were performed, deliberately changing temperature or rennet and CaCl₂ concentration. The comparison with fundamental rheology results confirmed the effectiveness of NIRS to monitor milk renneting. The reduced spectral acquisition time (10 s) showed data highly correlated (r > 0.99) to those acquired with longer acquisition time. The developed decision trees, based on PC1 scores and T² MSPC charts, confirmed the suitability of the proposed approach for the prediction of coagulation times and for the detection of possible failures. In conclusion, the work provides a robust but simple PAT approach to assist cheesemakers in monitoring the coagulation step in real-time.

PAT implementation for advanced process control in solid dosage manufacturing - A practical guide

The implementation of continuous pharmaceutical manufacturing requires advanced control strategies rather than traditional end product testing or an operation within a small range of controlled parameters. A high level of automation based on process models and hierarchical control concepts is desired. The relevant tools that have been developed and successfully tested in academic and industrial environments in recent years are now ready for utilization on the commercial scale. To date, the focus in Process Analytical Technology (PAT) has mainly been on achieving process understanding and quality control with the ultimate goal of real-time release testing (RTRT). This work describes the workflow for the development of an in-line monitoring strategy to support PAT-based real-time control actions and its integration into solid dosage manufacturing. All stages are discussed in this paper, from process analysis and definition of the monitoring task to technology assessment and selection, its process integration and the development of data acquisition.

Self-organizing maps-based generalized feature set selection for model adaption without reference data for batch process

When fourier transform infrared spectroscopy (FTIR) techniques combined with multivariate calibration are used to measure the key process features or analyte concentrations during batch process, model adaption is indispensable for maintaining the predictability of a primary calibration model in new secondary batches. Many model adaption methods conforming to the actual application scenario of batch process have been proposed. Here we report on a novel standard-free model adaption method without reference measurement called variable selection strategy with self-organizing maps (VSSOM). It uses self-organizing maps (SOM) to classify the whole spectral variables into multiple classes according to the spectra from primary batch and secondary batch, respectively; and the corresponding primary feature subsets and secondary feature subsets are formed firstly. Secondly, candidate feature subsets without empty elements are generated by operating intersection between any primary feature subsets and any secondary feature subsets. Thirdly, the candidate feature subset with minimum root mean square error of cross-validation (RMSECV) for the primary calibration set is selected as the optimal feature subset. In this manner, the optimal feature subset can be identified from the candidate feature subsets. In other words, VSSOM aims to create a stable and consistent feature subset across different batches provided that it selects better features within the intersection sets between primary feature subsets and any secondary feature subsets. Two batch process datasets (γ-polyglutamic acid fermentation and paeoniflorin extraction) are presented for comparing the VSSOM method with No transfer partial least squares (PLS), boxcar signal transfer (BST), successive projection algorithm (SPA), transfer component analysis (TCA) and domain-invariant iterative partial least squares (DIPALS). Experimental results show that VSSOM has superior performance and comparable prediction performance in all the scenarios.