Jakub Szlęk

Heuristic modeling of macromolecule release from PLGA microspheres

Dissolution of protein macromolecules from poly(lactic-co-glycolic acid) (PLGA) particles is a complex process and still not fully understood. As such, there are difficulties in obtaining a predictive model that could be of fundamental significance in design, development, and optimization for medical applications and toxicity evaluation of PLGA-based multiparticulate dosage form. In the present study, two models with comparable goodness of fit were proposed for the prediction of the macromolecule dissolution profile from PLGA micro- and nanoparticles. In both cases, heuristic techniques, such as artificial neural networks (ANNs), feature selection, and genetic programming were employed. Feature selection provided by fscaret package and sensitivity analysis performed by ANNs reduced the original input vector from a total of 300 input variables to 21, 17, 16, and eleven; to achieve a better insight into generalization error, two cut-off points for every method was proposed. The best ANNs model results were obtained by monotone multi-layer perceptron neural network (MON-MLP) networks with a root-mean-square error (RMSE) of 15.4, and the input vector consisted of eleven inputs. The complicated classical equation derived from a database consisting of 17 inputs was able to yield a better generalization error (RMSE) of 14.3. The equation was characterized by four parameters, thus feasible (applicable) to standard nonlinear regression techniques. Heuristic modeling led to the ANN model describing macromolecules release profiles from PLGA microspheres with good predictive efficiency. Moreover genetic programming technique resulted in classical equation with comparable predictability to the ANN model.

Analysis of pellet properties with use of artificial neural networks

The objective was to prepare neural models identifying relationships between formulation characteristics and pellet properties based on algorithmic approach of crucial variables selection and neuro-fuzzy systems application. The database consisted of information about 227 pellet formulations prepared by extrusion/spheronization method, with various model drugs and excipients. Cheminformatic description of excipients and model drugs was employed for numerical description of pellet formulations. Initial numbers of neural model inputs were up to around 3000. The inputs reduction procedure based on sensitivity analysis allowed to obtain less than 40 inputs for each model. The reduced models were subjects of fuzzy logic implementation resulting in logical rules tables providing human-readable rule sets applicable in future development of pellet formulations. Neural modeling enhanced knowledge about pelletization process and provided means for future computer-guided search for the optimal formulation.

Empirical modeling of the fine particle fraction for carrier-based pulmonary delivery formulations.

In vitro study of the deposition of drug particles is commonly used during development of formulations for pulmonary delivery. The assay is demanding, complex, and depends on: properties of the drug and carrier particles, including size, surface characteristics, and shape; interactions between the drug and carrier particles and assay conditions, including flow rate, type of inhaler, and impactor. The aerodynamic properties of an aerosol are measured in vitro using impactors and in most cases are presented as the fine particle fraction, which is a mass percentage of drug particles with an aerodynamic diameter below 5 μm. In the present study, a model in the form of a mathematical equation was developed for prediction of the fine particle fraction. The feature selection was performed using the R-environment package “fscaret”. The input vector was reduced from a total of 135 independent variables to 28. During the modeling stage, techniques like artificial neural networks, genetic programming, rule-based systems, and fuzzy logic systems were used. The 10-fold cross-validation technique was used to assess the generalization ability of the models created. The model obtained had good predictive ability, which was confirmed by a root-mean-square error and normalized root-mean-square error of 4.9 and 11%, respectively. Moreover, validation of the model using external experimental data was performed, and resulted in a root-mean-square error and normalized root-mean-square error of 3.8 and 8.6%, respectively.

Preformulation Studies on Solid Self-Emulsifying Systems in Powder Form Containing Magnesium Aluminometasilicate as Porous Carrier

The influence of alkaline and the neutral grade of magnesium aluminometasilicate as a porous solid carrier for the liquid self-emulsifying formulation with ibuprofen is investigated. Ibuprofen is dissolved in Labrasol, then this solution is adsorbed on the silicates. The drug to the silicate ratio is 1:2, 1:4, and 1:6, respectively. The properties of formulations obtained are analyzed, using morphological, porosity, crystallinity, and dissolution studies. Three solid self-emulsifying (S-SE) formulations containing Neusilin SG2 and six consisting of Neusilin US2 are in the form of powder without agglomerates. The nitrogen adsorption method shows that the solid carriers are mesoporous but they differ in a specific surface area, pore area, and the volume of pores. The adsorption of liquid SE formulation on solid silicate particles results in a decrease in their porosity. If the neutral grade of magnesium aluminometasilicate is used, the smallest pores, below 10 nm, are completely filled with liquid formulation, but there is still a certain number of pores of 40–100 nm. Dissolution studies of liquid SEDDS carried out in pH = 1.2 show that Labrasol improves the dissolution of ibuprofen as compared to the pure drug. Ibuprofen dissolution from liquid SE formulations examined in pH of 7.2 is immediate. The adsorption of the liquid onto the particles of the silicate causes a decrease in the amount of the drug released. Finally, more ibuprofen is dissolved from S-SE that consist of the neutral grade of magnesium aluminometasilicate than from the formulations containing the alkaline silicate.

Physicochemical Properties of Bosentan and Selected PDE-5 Inhibitors in the Design of Drugs for Rare Diseases.

The study provides the physicochemical characteristic of bosentan (BOS) in comparison to tadalafil (TA) and sildenafil citrate (SIL). Despite some reports dealing with thermal characteristic of SIL and TA, physicochemical properties of BOS have not been investigated so far. Recent clinical reports have indicated that the combination of bosentan and PDE-5 inhibitor can improve the effectiveness of pharmacotherapy of pulmonary arterial hypertension (PAH). However, in order to design personalized medicines for therapy of chronic rare diseases, detailed information on the thermal behaviour and solubility of each drug is indispensable. Thus, XRD, DSC and TGA-QMS analyses were applied to compare the properties of the drugs, their thermal stability as well as to identify the products of thermal degradation. The dehydration of BOS started at 70°C and was followed by the chemical degradation with the onset at 290°C. The highest thermal stability was stated for TA, which decomposed at ca. 320°C, whereas the lowest onset of the thermal decomposition process was stated for SIL, i.e. 190°C. The products of the drug decomposition were identified. FT-FIR was applied to study intra- and intermolecular interactions between the drug molecules. FT-MIR and Raman spectroscopy were used to examine the chemical structure of the drugs. Chemoinformatic tools were used to predict the polar surface area, pKa, or logP of the drugs. Their results were in line with solubility and dissolution studies.

Enhanced QSAR models for drug‐triggered inhibition of the main cardiac ion currents

The currently changing cardiac safety testing paradigm suggests, among other things, a shift towards using in silico models of cellular electrophysiology and assessment of a concomitant block of multiple ion channels. In this study, a set of four enhanced QSAR models have been developed: for the rapid delayed rectifying potassium current (IKr), slow delayed rectifying potassium current (IKs), peak sodium current (INa) and late calcium current (ICaL), predicting ion currents changes for the specific in vitro experiment from the 2D structure of the compounds. The models are a combination of both in vitro study parameters and physico‐chemical descriptors, which is a novel approach in drug–ion channels interactions modeling. Their predictive power assessed in the enhanced, more demanding than standard procedure, 10‐fold cross validation was reasonably high. Rough comparison with published pure in silico hERG interaction models shows that the quality of the model predictions does not differ from other models available in the public domain, however, it takes its advantage in accounting for inter‐experimental settings variability. Developed models are implemented in the Cardiac Safety Simulator, a commercially available platform enabling the in vitro–in vivo extrapolation of the drugs proarrhythmic effect and ECG simulation. A more comprehensive assessment of the effects of the compounds on ion channels allows for making more informed decisions regarding the risk – and thus avoidance – of exclusion of potentially safe and effective drugs. Copyright

PhEq_bootstrap: Open-Source Software for the Simulation of f2 Distribution in Cases of Large Variability in Dissolution Profiles

PhEq_bootstrap is a free software tool that uses the similarity factor (f2) to assess dissolution profile similarity in cases of large data variability. Its theoretical background is founded on bootstrapping, a statistical technique used to simulate the distribution of f2 values based on the available sample. It allows both justification of profile similarity and prospective simulations for the establishment of the formulation development endpoint. The software is FOSS (free open-source software) and is available online (1). INTRODUCTION Moore and Flanner (2) proposed simple measures for the distance between the two curves representing dissolution profiles of two dosage forms, namely the difference (f1) and similarity (f2) factors. Because of its mathematical simplicity and lack of mechanistic assumptions, the similarity factor (f2) is currently one of the most commonly used methods for the comparison of dissolution profiles (3). Although it is a simple method, f2 is restricted to the following conditions (4): • Minimum of three points in the profile. • Twelve units for each reference and test product. • No more than one point over 85%. • RSD for dissolution points less than 10% (first point less than 20%). The appropriate construction of the analytical protocol ensures the number and the range of the dissolution points applicable to the f2 computation. However, the variability of the dissolution points is attributed mainly to the immanent characteristics of the dosage form (i.e., API solubility rate variations, coating non-uniformity, etc.). Thus, variability is usually beyond the control of the analyst, and its effect has to be minimized by appropriate statistical techniques to compare dissolution profiles with maximum reliability. Two major groups of techniques applicable here are based on covariance matrix and bootstrap. Analysis of the covariance matrix is used in the Mahalanobis distance technique of the direct profile comparison. Since it is a multivariate technique of profile comparison, f2 is not used here. The Mahalanobis distance technique suffers from the restrictions of the matrix inversion techniques, which are sensitive to the covariance matrix characteristics; thus, it is sometimes unable to handle real data. Moreover, interpretation of the Mahalonobis distance requires complex computations of the confidence intervals. Therefore, this approach is not as versatile and simple to interpret as f2. In contrast to the Mahalanobis distance technique, the bootstrap technique allows the use of f2; however, it is also used not as a point estimator but as a confidence interval. The bootstrap technique is mathematically and algorithmically much simpler than the Mahalanobis distance technique. Moreover, it is easier to interpret based on the use of commonly known rules describing f2 values and their meaning for the decision on the similarity of the analyzed profiles. Based on the above considerations, the aim of this work was to create a computer program able to provide analysis of the f2 confidence intervals with the bootstrap technique to help in the assessment of the similarity between dissolution profiles in cases of large variability in the dissolution data. PhEq_Bootstrap FEATURES AND THEORETICAL BACKGROUND PhEq_bootstrap is software that handles the above presented restrictions of f2 concerning RSD of the dissolution points. The bootstrap technique is used to simulate f2 distribution to assess the worst-case scenario as a lower confidence interval of the expected values of f2. Computations of expected values and unbiased estimators of f2 are based on the publication of Shah et al. (5), where mathematical proof of this concept is presented and discussed. To carry out the computations, a new population of dissolution profiles was generated numerically by the bootstrap technique, where the new samples were the result of random sampling with replacement of the dissolution profiles. This procedure was performed for both the reference and test profiles. Two modes of the sampling proce*Corresponding author. e-mail: mfmendyk@cyf-kr.edu.pl dx.doi.org/10.14227/DT200113P13

Effect of roll compaction on granule size distribution of microcrystalline cellulose–mannitol mixtures: computational intelligence modeling and parametric analysis

Dry granulation using roll compaction is a typical unit operation for producing solid dosage forms in the pharmaceutical industry. Dry granulation is commonly used if the powder mixture is sensitive to heat and moisture and has poor flow properties. The output of roll compaction is compacted ribbons that exhibit different properties based on the adjusted process parameters. These ribbons are then milled into granules and finally compressed into tablets. The properties of the ribbons directly affect the granule size distribution (GSD) and the quality of final products; thus, it is imperative to study the effect of roll compaction process parameters on GSD. The understanding of how the roll compactor process parameters and material properties interact with each other will allow accurate control of the process, leading to the implementation of quality by design practices. Computational intelligence (CI) methods have a great potential for being used within the scope of quality by design approach. The main objective of this study was to show how the computational intelligence techniques can be useful to predict the GSD by using different process conditions of roll compaction and material properties. Different techniques such as multiple linear regression, artificial neural networks, random forest, Cubist and k-nearest neighbors algorithm assisted by sevenfold cross-validation were used to present generalized models for the prediction of GSD based on roll compaction process setting and material properties. The normalized root-mean-squared error and the coefficient of determination (R2) were used for model assessment. The best fit was obtained by Cubist model (normalized root-mean-squared error =3.22%, R2=0.95). Based on the results, it was confirmed that the material properties (true density) followed by compaction force have the most significant effect on GSD.

From Black-Box to Transparent Computational Intelligence Models: A Pharmaceutical Case Study

Pharmaceutical Industry is tightly regulated owing to health concerns. Over the years, the use of computational intelligence (CI) tools has increased in the pharmaceutical research and development, manufacturing, and quality control. CI models for tensile strength of tablets based on the formulation design and manufacturing parameters have been established. Best models exhibit NRMSE of 7.2%. Implicitly, CI tools work in a black -- box fashion which makes it difficult to inspect the model. Owing to quality and safety concerns, it is imperative for the pharmaceutical industry and the regulatory authorities to have knowledge of how the CI models work. This work uses data from a galencial tableting study to establish models for the outcome of tensile strength from various CI techniques and makes an attempt to make the models as transparent as possible. Tree based ensembles and symbolic regression methods are presented as transparent models with extracted rules and mathematical formula, respectively, explaining the CI models in greater detail.

Computational intelligence models to predict porosity of tablets using minimum features

The effects of different formulations and manufacturing process conditions on the physical properties of a solid dosage form are of importance to the pharmaceutical industry. It is vital to have in-depth understanding of the material properties and governing parameters of its processes in response to different formulations. Understanding the mentioned aspects will allow tighter control of the process, leading to implementation of quality-by-design (QbD) practices. Computational intelligence (CI) offers an opportunity to create empirical models that can be used to describe the system and predict future outcomes in silico. CI models can help explore the behavior of input parameters, unlocking deeper understanding of the system. This research endeavor presents CI models to predict the porosity of tablets created by roll-compacted binary mixtures, which were milled and compacted under systematically varying conditions. CI models were created using tree-based methods, artificial neural networks (ANNs), and symbolic regression trained on an experimental data set and screened using root-mean-square error (RMSE) scores. The experimental data were composed of proportion of microcrystalline cellulose (MCC) (in percentage), granule size fraction (in micrometers), and die compaction force (in kilonewtons) as inputs and porosity as an output. The resulting models show impressive generalization ability, with ANNs (normalized root-mean-square error [NRMSE] =1%) and symbolic regression (NRMSE =4%) as the best-performing methods, also exhibiting reliable predictive behavior when presented with a challenging external validation data set (best achieved symbolic regression: NRMSE =3%). Symbolic regression demonstrates the transition from the black box modeling paradigm to more transparent predictive models. Predictive performance and feature selection behavior of CI models hints at the most important variables within this factor space.