86 results found
Odgers J, Kappatou C, Misener R, et al., 2023, Probabilistic predictions for partial least squares using bootstrap, AIChE Journal, Vol: 69, Pages: 1-16, ISSN: 0001-1541
Modeling the uncertainty in partial least squares (PLS) is made difficult because of the nonlinear effect of the observed data on the latent space that the method finds. We present an approach, based on bootstrapping, that automatically accounts for these nonlinearities in the parameter uncertainty, allowing us to equally well represent confidence intervals for points lying close to or far away from the latent space. To show the opportunities of this approach, we develop applications in determining the Design Space for industrial processes and model the uncertainty of spectroscopy data. Our results show the benefits of our method for accounting for uncertainty far from the latent space for the purposes of Design Space identification, and match the performance of well established methods for spectroscopy data.
Nașcu I, Diangelakis NA, Muñoz SG, et al., 2023, Advanced model predictive control strategies for evaporation processes in the pharmaceutical industries, Computers and Chemical Engineering, Vol: 173, ISSN: 0098-1354
In this paper we present a framework to design control systems for an evaporation process in the pharmaceutical industry with the aim to deliver guaranteed operability for different molecules and under different thermodynamic scenarios. Based on a mathematical model developed within the gPROMS platform calibrated and validated with real data from experiments, three control methods are implemented and compared, (i) Proportional Integrative Derivative control (PID), (ii) Model Predictive Control (MPC) and (iii) explicit/multi-parametric Model Predictive Control (mp-MPC). The performance and limits of the derived control schemes are then established and tested for reference tracking as well as disturbances rejection.
Kappatou C, Odgers J, García-Muñoz S, et al., 2023, An optimization approach coupling pre-processing with model regression for enhanced chemometrics, Industrial and Engineering Chemistry Research, Vol: 62, Pages: 6196-6213, ISSN: 0888-5885
Chemometric methods are broadly used in the chemical and biochemical sectors. Typically, derivation of a regression model follows data preprocessing in a sequential manner. Yet, preprocessing can significantly influence the regression model and eventually its predictive ability. In this work, we investigate the coupling of preprocessing and model parameter estimation by incorporating them simultaneously in an optimization step. Common model selection techniques rely almost exclusively on the performance of some accuracy metric, yet having a quantitative metric for model robustness can prolong model up-time. Our approach is applied to optimize for model accuracy and robustness. This requires the introduction of a novel mathematical definition for robustness. We test our method in a simulated set up and with industrial case studies from multivariate calibration. The results highlight the importance of both accuracy and robustness properties and illustrate the potential of the proposed optimization approach toward automating the generation of efficient chemometric models.
Zhao F, Ochoa MP, Grossmann IE, et al., 2022, Novel formulations of flexibility index and design centering for design space definition, COMPUTERS & CHEMICAL ENGINEERING, Vol: 166, ISSN: 0098-1354
Kusumo K, Kuriyan K, Vaidyaraman S, et al., 2022, Probabilistic framework for optimal experimental campaigns in the presence of operational constraints, Reaction Chemistry and Engineering, Vol: 7, Pages: 2359-2374, ISSN: 2058-9883
The predictive capability of any mathematical model is intertwined with the quality of experimentaldata collected for its calibration. Model-based design of experiments helps compute maximallyinformative campaigns for model calibration. But in early stages of model development it is crucial toaccount for model uncertainties to mitigate the risk of uninformative or infeasible experiments. Thisarticle presents a new method to design optimal experimental campaigns subject to hard constraintsunder uncertainty, alongside a tractable computational framework. This computational frameworkinvolves two stages, whereby the feasible experimental space is sampled using a probabilistic approachin the first stage, and a continuous-effort optimal experiment design is determined by searching overthe sampled feasible space in the second stage. The tractability of this methodology is demonstratedon a case study involving the exothermic esterification of priopionic anhydride with significant risk ofthermal runaway during experimentation. An implementation is made freely available based on thePython packages DEUS and Pydex.
Johnson BJ, Sen M, Hanson J, et al., 2022, Stochastic analysis and modeling of pharmaceutical screw feeder mass flow rates, INTERNATIONAL JOURNAL OF PHARMACEUTICS, Vol: 621, ISSN: 0378-5173
Zhao F, Grossmann IE, Garcia-Munoz S, et al., 2022, Design space description through adaptive sampling and symbolic computation, AICHE JOURNAL, Vol: 68, ISSN: 0001-1541
Kusumo K, Kuriyan K, Vaidyaraman S, et al., 2022, Risk mitigation in model-based experiment design: a continuous-effort approach to optimal campaigns, Computers and Chemical Engineering, Vol: 159, ISSN: 0098-1354
A key challenge in maximizing the effectiveness of model-based design of experiments for calibrating nonlinear process models is the inaccurate prediction of information that is afforded by each new experiment. We present a novel methodology to exploit prior probability distributions of model parameter estimates in a bi-objective optimization formulation, where a conditional-value-at-risk criterion is considered alongside an average information criterion. We implement a tractable numerical approach that discretizes the experimental design space and leverages the concept of continuous-effort experimental designs in a convex optimization formulation. We demonstrate effectiveness and tractability through three case studies, including the design of dynamic experiments. In one case, the Pareto frontier comprises experimental campaigns that significantly increase the information content in the worst-case scenarios. In another case, the same campaign is proven to be optimal irrespective of the risk attitude. An open-source implementation of the methodology is made available in the Python software Pydex.
Sen M, Arguelles AJ, Stamatis SD, et al., 2021, An optimization-based model discrimination framework for selecting an appropriate reaction kinetic model structure during early phase pharmaceutical process development, REACTION CHEMISTRY & ENGINEERING, Vol: 6, Pages: 2092-2103, ISSN: 2058-9883
Destro F, García Muñoz S, Bezzo F, et al., 2021, Powder composition monitoring in continuous pharmaceutical solid-dosage form manufacturing using state estimation - Proof of concept., Int J Pharm, Vol: 605
In continuous solid-dosage form manufacturing, the powder feeding system is responsible for supplying downstream the correct formulation of the drug product ingredients. The composition of the powder delivered by the feeding system is inferred from the measurements of powder mass flow from the system feeders. The mass flows are, in turn, inferred from the loss in weight measured in the feeder hoppers. Most loss-in-weight feeders post-process the mass flow signal to deliver a smoothed value to the user. However, such estimated mass flows can exhibit a low signal-to-noise ratio. As the feeders are critical elements of the control strategy of the manufacturing line, better instantaneous estimates of mass flow are desirable for improving the quality assurance. In this study, we propose a model-based approach for monitoring the composition of the powder fed to a continuous solid-dosage line. The monitoring system is based on a moving-horizon state estimator, which carries out model-based reconciliation of the feeder mass measurements, thus enabling accurate composition estimation of the powder mixture. Experimental datasets from a direct compression line are used to validate the methodology. Results demonstrate improvement with respect to current industrial solutions.
Wang H, Dieringer J, Guntz S, et al., 2021, Portfolio-Wide Optimization of Pharmaceutical R&D Activities Using Mathematical Programming, INFORMS JOURNAL ON APPLIED ANALYTICS, Vol: 51, Pages: 262-279, ISSN: 2644-0865
Ochoa MP, Garcia-Munoz S, Stamatis S, et al., 2021, Novel flexibility index formulations for the selection of the operating range within a design space, COMPUTERS & CHEMICAL ENGINEERING, Vol: 149, ISSN: 0098-1354
Sen M, Garcia Munoz S, 2021, Development and implementation of a hybrid scale up model for a batch high shear wet granulation operation, AICHE JOURNAL, Vol: 67, ISSN: 0001-1541
Zhao F, Grossmann IE, Garcia-Munoz S, et al., 2021, Flexibility index of black-box models with parameter uncertainty through derivative-free optimization, AICHE JOURNAL, Vol: 67, ISSN: 0001-1541
Kusumo KP, Kuriyan K, García-Muñoz S, et al., 2021, Continuous-Effort Approach to Model-Based Experimental Designs, Computer Aided Chemical Engineering, Pages: 867-873
Model-based design of experiments is a technique for accelerating the development of mathematical models. Through maximally informative experiments, time and resources for estimating uncertain model parameters are minimized. This article presents a method for computing effort-based experimental designs, whereby designs are akin to experimental recipes. As well as identifying which experiments are the most informative, the optimal experimental effort to dedicate to each experiment is also optimized. Upon discretizing the experimental design space and treating the efforts as continuous decision variables, this method leads to convex optimization problems regardless of the model structure, which is ideal for large, parallel experimental campaigns. The case study of a batch reactor model with four parameters is presented to illustrate the methodology.
Short M, Biegler LT, Garcia-Munoz S, et al., 2020, Estimating variances and kinetic parameters from spectra across multiple datasets using KIPET, CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, Vol: 203, ISSN: 0169-7439
Christodoulou C, Sorensen E, Khair AS, et al., 2020, A model for the fluid dynamic behavior of a film coating suspension during tablet coating, CHEMICAL ENGINEERING RESEARCH & DESIGN, Vol: 160, Pages: 301-320, ISSN: 0263-8762
Destro F, Facco P, García Muñoz S, et al., 2020, A hybrid framework for process monitoring: Enhancing data-driven methodologies with state and parameter estimation, Journal of Process Control, Vol: 92, Pages: 333-351, ISSN: 0959-1524
In this study we bridge traditional standalone data-driven and knowledge-driven process monitoring approaches by proposing a novel hybrid framework that exploits the advantages of both simultaneously. Namely, we design a process monitoring system based on a data-driven model that includes two different data types: i) “actual” data coming from sensor measurements, and ii) “virtual” data coming from a state estimator, based on a first-principles model of the system under investigation. We test the proposed approach on two simulated case studies: a continuous polycondensation process for the synthesis of poly-ethylene terephthalate, and a fed-batch fermentation process for the manufacturing of penicillin. The hybrid monitoring model shows superior fault detection and diagnosis performances with respect to conventional monitoring techniques, even when the first-principles model is relatively simple and process/model mismatch exists.
Garcia Munoz S, Hernandez Torres E, 2020, Supervised Extended Iterative Optimization Technology for Estimation of Powder Compositions in Pharmaceutical Applications: Method and Lifecycle Management, INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH, Vol: 59, Pages: 10072-10081, ISSN: 0888-5885
Short M, Thierry D, Rodriguez S, et al., 2020, salvadorgarciamunoz/kipet: Stable release with multiple experimental datasets
New KIPET Version 1.1.0!New stable release, which includes the features and code of the latest KIPET publication.Major Features Multiple Experimental DatasetsThe major addition to the KIPET package since the last release is the inclusion of the MultipleExperimentsEstimator class that allows for multiple experimental datasets to be analyzed simultaneously. The class allows for different models or the same model to be inputted, with parameters that are local or global to the specific datasets automatically detected. The class contains functions to run separate variance estimation upon each dataset that can also be used to initialize the problem. In addition the parameter estimation can be done simultaneously across multiple datasets and models for both spectra and concentration problems with or without shared spectra between species.New Variance Estimation MethodThe new variance estimation method described in the second KIPET paper is also included, providing a more rigorous approach to the variance estimation.Minor FeaturesA host of new examples to showcase the multiple experiments estimation, new documentation, as well as a few minor bug fixes.
Bascone D, Galvanin F, Shah N, et al., 2020, Hybrid mechanistic-empirical approach to the modeling of twin screw feeders for continuous tablet manufacturing, Industrial and Engineering Chemistry Research, Vol: 59, Pages: 6650-6661, ISSN: 0888-5885
Nowadays, screw feeders are popular equipment in the pharmaceutical industry. However, despite the increasing research in the last decade in the manufacturing of powder-based products, there is still a lack of knowledge on the physics governing the dynamic behavior of these systems. As a result, data-driven models have often been used to address process design, optimization, and control applications. In this paper, a methodology for the modeling of twin screw feeders has been suggested. A first order plus dead time model has been developed, where a hybrid mechanistic-empirical approach has been used. Different powders and two screw feeder geometries have been investigated. The model predictions are in good agreement with the experimental measurements when the 35 mm diameter screws are employed. When the 20 mm diameter screws are used, the validity range of the model is limited for the least cohesive powders, suggesting that their screw speed-dependent resistance to flow in small screws requires further investigations.
Schenk C, Short M, Rodriguez JS, et al., 2020, Introducing KIPET: A novel open-source software package for kinetic parameter estimation from experimental datasets including spectra, COMPUTERS & CHEMICAL ENGINEERING, Vol: 134, ISSN: 0098-1354
Kucherenko S, Giamalakis D, Shah N, et al., 2020, Computationally efficient identification of probabilistic design spaces through application of metamodeling and adaptive sampling, Computers & Chemical Engineering, Vol: 132, Pages: 1-9, ISSN: 0098-1354
The design space (DS) is defined as the combination of materials and process conditions which provides assurance of quality for a pharmaceutical product (e.g. purity, potency, uniformity). A model-based approach to identify a probability-based design space requires simulations across the entire process parameter space (certain) and the uncertain model parameter space and material properties space if explicitly considered by the model. This exercise is a demanding task. A novel theoretical and numerical framework for determining probabilistic DS using metamodelling and adaptive sampling is developed. Several approaches were proposed and tested among which the most efficient is a new multi-step adaptive technique based using a metamodel for a probability map as an acceptance-rejection criterion to optimize sampling to identify the DS. It is shown that application of metamodel-based filters can significantly reduce model complexity and computational costs with speed up of two orders of magnitude observed here.
Christodoulou C, Sorensen E, Garcia-Munoz S, et al., 2020, Mathematical Modeling of Spray Impingement and Film Formation on Pharmaceutical Tablets during Coating, CHEMICAL ENGINEERING RESEARCH & DESIGN, Vol: 153, Pages: 768-788, ISSN: 0263-8762
Kusumo KP, Gomoescu L, Paulen R, et al., 2020, Nested Sampling Strategy for Bayesian Design Space Characterization, Editors: Pierucci, Manenti, Bozzano, Manca, Publisher: ELSEVIER SCIENCE BV, Pages: 1957-1962
Kusumo KP, Gomoescu L, Paulen R, et al., 2019, Bayesian approach to probabilistic design space characterization: a nested sampling strategy, Industrial & Engineering Chemistry Research, Vol: 59, Pages: 2396-2408, ISSN: 0888-5885
Quality by design in pharmaceutical manufacturing hinges on computational methods and tools that are capable of accurate quantitative prediction of the design space. This paper investigates Bayesian approaches to design space characterization, which determine a feasibility probability that can be used as a measure of reliability and risk by the practitioner. An adaptation of nested sampling—a Monte Carlo technique introduced to compute Bayesian evidence—is presented. The nested sampling algorithm maintains a given set of live points through regions with increasing probability feasibility until reaching a desired reliability level. It furthermore leverages efficient strategies from Bayesian statistics for generating replacement proposals during the search. Features and advantages of this algorithm are demonstrated by means of a simple numerical example and two industrial case studies. It is shown that nested sampling can outperform conventional Monte Carlo sampling and be competitive with flexibility-based optimization techniques in low-dimensional design space problems. Practical aspects of exploiting the sampled design space to reconstruct a feasibility probability map using machine learning techniques are also discussed and illustrated. Finally, the effectiveness of nested sampling is demonstrated on a higher-dimensional problem, in the presence of a complex dynamic model and significant model uncertainty.
Chen W, Biegler LT, Muñoz SG, 2019, A Unified Framework for Kinetic Parameter Estimation Based on Spectroscopic Data with or without Unwanted Contributions, Industrial and Engineering Chemistry Research, Vol: 58, Pages: 13651-13663, ISSN: 0888-5885
Kinetic estimation of chemical reactions is considered to be a crucial step prior to design of a robust, controllable, and safe production process. Spectroscopic measurements are widely used for kinetic analysis. However, there may be unwanted contributions in the measured spectra, which may come from the instrumental variations (such as the baseline shift or distortion) or from the presence of inert absorbing interferences with no kinetic behavior. Here, the kinetic estimation problem with time-invariant contributions is studied in depth, and we derive conditions where the estimation accuracy is not affected by the time-invariant contributions. Moreover, kinetic parameter estimation and separation of time-invariant contributions can be performed simultaneously under proper conditions. Also, an approach for kinetic parameter estimation based on spectra with time-variant contributions is proposed. Finally, a novel unified framework is developed for kinetic parameter estimation when there is no prior information for unwanted contributions.
Borhani T, Garcia-Munoz S, Luciani C, et al., 2019, Hybrid QSPR models for the prediction of the free energy of solvation of organic solute/solvent pairs, Physical Chemistry Chemical Physics, Vol: 21, Pages: 13706-13720, ISSN: 1463-9076
Due to the importance of the Gibbs free energy of solvation in understanding many physicochemical phenomena, including lipophilicity, phase equilibria and liquid-phase reaction equilibrium and kinetics, there is a need for predictive models that can be applied across large sets of solvents and solutes. In this paper, we propose two quantitative structure property relationships (QSPRs) to predict the Gibbs free energy of solvation, developed using partial least squares (PLS) and multivariate linear regression (MLR) methods for 295 solutes in 210 solvents with total number of data points of 1777. Unlike other QSPR models, the proposed models are not restricted to a specific solvent or solute. Furthermore, while most QSPR models include either experimental or quantum mechanical descriptors, the proposed models combine both, using experimental descriptors to represent the solvent and quantum mechanical descriptors to represent the solute. Up to twelve experimental descriptors and nine quantum mechanical descriptors are considered in the proposed models. Extensive internal and external validation is undertaken to assess model accuracy s in predicting the Gibbs free energy of solvation for a large number of solute/solvent pairs. The best MLR model, which includes three solute descriptors and two solvent properties, yields a coefficient of determination (R2) of 0.88 and a root mean squared error (RMSE) of 0.59 kcal/mol for the training set. The best PLS model includes six latent variables, and has a R2 value of 0.91 and a RMSE of 0.52 kcal/mol. The proposed models are compared to selected results based on continuum solvation quantum chemistry calculations. They enable the fast prediction of the Gibbs free energy of solvation of a wide range of solutes in different solvents.
Reizman BJ, Burt JL, Frank SA, et al., 2019, Data-Driven Prediction of Risk in Drug Substance Starting Materials, ORGANIC PROCESS RESEARCH & DEVELOPMENT, Vol: 23, Pages: 1429-1441, ISSN: 1083-6160
This chapter discusses modeling techniques applied to drug product formulation and processing operations as support to the design, development, and scale-up for solid oral drug products. It discusses these process modeling techniques with case studies ranging from raw material specifications to process parameter predictions. In general, the main unit operations utilized to produce tablets or capsules include: blending and other powder processing, dry or wet granulation, tablet compression (powder compaction) or encapsulation, lubrication final blending, and possibly film coating. Specific consideration is given in the chapter to transfer and scale-up issues along with general process design-related challenges to pharmaceutical process research and development. The chapter illustrates how the thermodynamic film coating model can be used to predict specific operating conditions, assist with scale-up from one coater to another, and also simulate different operation scenarios on a given film coater.
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.