SerotoninAI: Serotonergic System Focused, Artificial Intelligence-Based Application for Drug Discovery

SerotoninAI is an innovative web application for scientific purposes focused on the serotonergic system. By leveraging SerotoninAI, researchers can assess the affinity (pKi value) of a molecule to all main serotonin receptors and serotonin transporters based on molecule structure introduced as SMILES. Additionally, the application provides essential insights into critical attributes of potential drugs such as blood-brain barrier penetration and human intestinal absorption. The complexity of the serotonergic system demands advanced tools for accurate predictions, which is a fundamental requirement in drug development. SerotoninAI addresses this need by providing an intuitive user interface that generates predictions of pKi values for the main serotonergic targets. The application is freely available on the Internet at https://serotoninai.streamlit.app/, implemented in Streamlit with all major web browsers supported. Currently, to the best of our knowledge, there is no tool that allows users to access affinity predictions for serotonergic targets without registration or financial obligations. SerotoninAI significantly increases the scope of drug development activities worldwide. The source code of the application is available at https://github.com/nczub/SerotoninAI_streamlit.

Investigating the Impact of Co-processed Excipients on the Formulation of Bromhexine Hydrochloride Orally Disintegrating Tablets (ODTs)

PURPOSE

Orodispersible tablets (orally disintegrating tablets, ODTs) have been used in pharmacotherapy for over 20 years since they overcome the problems with swallowing solid dosage forms. The successful formula manufactured by direct compression shall ensure acceptable mechanical strength and short disintegration time. Our research aimed to develop ODTs containing bromhexine hydrochloride suitable for registration in accordance with EMA requirements.

METHODS

We examined the performance of five multifunctional co-processed excipients, i.e., F-Melt® C, F-Melt® M, Ludiflash®, Pharmaburst® 500 and Prosolv® ODT G2 as well as self-prepared physical blend of directly compressible excipients. We tested powder flow, true density, compaction characteristics and tableting speed sensitivity.

RESULTS

The manufacturability studies confirmed that all the co-processed excipients are very effective as the ODT formula constituents. We noticed superior properties of both F-Melt's®, expressed by good mechanical strength of tablets and short disintegration time. Ludiflash® showed excellent performance due to low works of plastic deformation, elastic recovery and ejection. However, the tablets released less than 30% of the drug. Also, the self-prepared blend of excipients was found sufficient for ODT application and successfully transferred to production scale. Outcome of the scale-up trial revealed that the tablets complied with compendial requirements for orodispersible tablets.

CONCLUSIONS

We proved that the active ingredient cannot be absorbed in oral cavity and its dissolution profiles in media representing upper part of gastrointestinal tract are similar to marketed immediate release drug product. In our opinion, the developed formula is suitable for registration within the well-established use procedure without necessity of bioequivalence testing.

Application of automated machine learning in the identification of multi-target-directed ligands blocking PDE4B, PDE8A, and TRPA1 with potential use in the treatment of asthma and COPD

The aim of the study was to develop AutoML models to search for novel MTDL chemotypes blocking PDE4B, PDE8A, and TRPA1. Asthma and COPD are characterized by complex pathophysiology associated with chronic inflammation, bronchoconstriction, and bronchial hyperresponsiveness resulting in airway remodeling. A possible comprehensive solution that could fully counteract the pathological processes of both diseases are rationally designed multi-target-directed ligands (MTDLs), combining PDE4B and PDE8A inhibition with TRPA1 blockade. Regression models were developed for each of the biological targets using "mljar-supervised". On their basis, virtual screenings of commercially available compounds derived from the ZINC15 database were performed. A common group of compounds placed within the top results was selected as potential novel chemotypes of multifunctional ligands. This study represents the first attempt to discover the potential MTDLs inhibiting three biological targets. The obtained results prove the usefulness of AutoML methodology in the identification of hits from the big compound databases.

Artificial Intelligence-Based Quantitative Structure-Property Relationship Model for Predicting Human Intestinal Absorption of Compounds with Serotonergic Activity

Oral medicines represent the largest pharmaceutical market area. To achieve a therapeutic effect, a drug must penetrate the intestinal walls, the main absorption site for orally delivered active pharmaceutical ingredients (APIs). Indeed, predicting drug absorption can facilitate candidate screening and reduce time to market. Algorithms are available with good prediction accuracy that however focus only on solubility. In this work, we focused on drug permeability looking at human intestinal absorption as a marker for intestinal bioavailability. Being of considerable therapeutic relevance, APIs with serotonergic activity were selected as a dataset. Due to process complexity, experimental data scarcity, and variability, we turned toward an artificial intelligence (AI)-based system, which is a hierarchical combination of classification and regression models. This combination of seemingly two models into a single system widens the space of molecules classified as highly permeable with high accuracy. The specialized and optimized system enables in silico and structure-based prediction with a high degree of certainty. Predictions in external validation allowed correct selection of the 38% of highly permeable molecules without any false positives. The proposed system based on AI represents a promising tool useful for oral drug screening at an early stage of drug discovery and development. Datasets and the obtained models are available on the GitHub platform (https://github.com/nczub/HIA_5-HT).

Formulation of Dosage Forms with Proton Pump Inhibitors: State of the Art, Challenges and Future Perspectives

Since their introduction to pharmacotherapy, proton pump inhibitors (PPIs) have been widely used in the treatment of numerous diseases manifested by excessive secretion of gastric acid. Despite that, there are still unmet needs regarding their availability for patients of all age groups. Their poor stability hinders the development of formulations in which dose can be easily adjusted. The aim of this review is to describe the discovery and development of PPIs, discuss formulation issues, and present the contemporary solutions, possibilities, and challenges in formulation development. The review outlines the physicochemical characteristics of PPIs, connects them with pharmacokinetic and pharmacodynamic properties, and describes the stability of PPIs, including the identification of the most important factors affecting them. Moreover, the possibilities for qualitative and quantitative analysis of PPIs are briefly depicted. This review also characterizes commercial preparations with PPIs available in the US and EU. The major part of the review is focused on the presentation of the state of the art in the development of novel formulations with PPIs covering various approaches employed in this process: nanoparticles, microparticles, minitablets, pellets, bilayer, floating, and mucoadhesive tablets, as well as parenteral, transdermal, and rectal preparations. It also anticipates further possibilities in the development of PPIs dosage forms. It is especially addressed to the researchers developing new formulations containing PPIs, since it covers the most important formulary issues that need to be considered before a decision on the selection of the formula is made. It may help in avoiding unnecessary efforts in this process and choosing the best approach. The review also presents an up-to-date database of publications focused on the pharmaceutical technology of formulations with PPIs.

Do AutoML-Based QSAR Models Fulfill OECD Principles for Regulatory Assessment? A 5-HT1A Receptor Case

The drug discovery and development process requires a lot of time, financial, and workforce resources. Any reduction in these burdens might benefit all stakeholders in the healthcare domain, including patients, government, and companies. One of the critical stages in drug discovery is a selection of molecular structures with a strong affinity to a particular molecular target. The possible solution is the development of predictive models and their application in the screening process, but due to the complexity of the problem, simple and statistical models might not be sufficient for practical application. The manuscript presents the best-in-class predictive model for the serotonin 1A receptor affinity and its validation according to the Organization for Economic Co-operation and Development guidelines for regulatory purposes. The model was developed based on a database with close to 9500 molecules by using an automatic machine learning tool (AutoML). The model selection was conducted based on the Akaike information criterion value and 10-fold cross-validation routine, and later good predictive ability was confirmed with an additional external validation dataset with over 700 molecules. Moreover, the multi-start technique was applied to test if an automatic model development procedure results in reliable results.

Curated Database and Preliminary AutoML QSAR Model for 5-HT1A Receptor

Introduction of a new drug to the market is a challenging and resource-consuming process. Predictive models developed with the use of artificial intelligence could be the solution to the growing need for an efficient tool which brings practical and knowledge benefits, but requires a large amount of high-quality data. The aim of our project was to develop quantitative structure–activity relationship (QSAR) model predicting serotonergic activity toward the 5-HT1A receptor on the basis of a created database. The dataset was obtained using ZINC and ChEMBL databases. It contained 9440 unique compounds, yielding the largest available database of 5-HT1A ligands with specified pKi value to date. Furthermore, the predictive model was developed using automated machine learning (AutoML) methods. According to the 10-fold cross-validation (10-CV) testing procedure, the root-mean-squared error (RMSE) was 0.5437, and the coefficient of determination (R²) was 0.74. Moreover, the Shapley Additive Explanations method (SHAP) was applied to assess a more in-depth understanding of the influence of variables on the model’s predictions. According to to the problem definition, the developed model can efficiently predict the affinity value for new molecules toward the 5-HT1A receptor on the basis of their structure encoded in the form of molecular descriptors. Usage of this model in screening processes can significantly improve the process of discovery of new drugs in the field of mental diseases and anticancer therapy.

Data-Driven Modeling of the Bicalutamide Dissolution from Powder Systems

Low solubility of active pharmaceutical compounds (APIs) remains an important challenge in dosage form development process. In the manuscript, empirical models were developed and analyzed in order to predict dissolution of bicalutamide (BCL) from solid dispersion with various carriers. BCL was chosen as an example of a poor water-soluble API. Two separate datasets were created: one from literature data and another based on in-house experimental data. Computational experiments were conducted using artificial intelligence tools based on machine learning (AI/ML) with a plethora of techniques including artificial neural networks, decision trees, rule-based systems, and evolutionary computations. The latter resulting in classical mathematical equations provided models characterized by the lowest prediction error. In-house data turned out to be more homogeneous, as well as formulations were more extensively characterized than literature-based data. Thus, in-house data resulted in better models than literature-based data set. Among the other covariates, the best model uses for prediction of BCL dissolution profile the transmittance from IR spectrum at 1260 cm^-1 wavenumber. Ab initio modeling-based in silico simulations were conducted to reveal potential BCL-excipients interaction. All crucial variables were selected automatically by AI/ML tools and resulted in reasonably simple and yet predictive models suitable for application in Quality by Design (QbD) approaches. Presented data-driven model development using AI/ML could be useful in various problems in the field of pharmaceutical technology, resulting in both predictive and investigational tools revealing new knowledge.

Physiologically based pharmacokinetic-quantitative systems toxicology and safety (PBPK-QSTS) modeling approach applied to predict the variability of amitriptyline pharmacokinetics and cardiac safety in populations and in individuals

The physiologically based pharmacokinetic (PBPK) models allow for predictive assessment of variability in population of interest. One of the future application of PBPK modeling is in the field of precision dosing and personalized medicine. The aim of the study was to develop PBPK model for amitriptyline given orally, predict the variability of cardiac concentrations of amitriptyline and its main metabolite-nortriptyline in populations as well as individuals, and simulate the influence of those xenobiotics in therapeutic and supratherapeutic concentrations on human electrophysiology. The cardiac effect with regard to QT and RR interval lengths was assessed. The Emax model to describe the relationship between amitriptyline concentration and heart rate (RR) length was proposed. The developed PBPK model was used to mimic 29 clinical trials and 19 cases of amitriptyline intoxication. Three clinical trials and 18 cases were simulated with the use of PBPK-QSTS approach, confirming lack of cardiotoxic effect of amitriptyline in therapeutic doses and the increase in heart rate along with potential for arrhythmia development in case of amitriptyline overdose. The results of our study support the validity and feasibility of the PBPK-QSTS modeling development for personalized medicine.

Mechanistic Physiologically Based Pharmacokinetic (PBPK) Model of the Heart Accounting for Inter-Individual Variability: Development and Performance Verification

Modern model-based approaches to cardiac safety and efficacy assessment require accurate drug concentration-effect relationship establishment. Thus, knowledge of the active concentration of drugs in heart tissue is desirable along with inter-subject variability influence estimation. To that end, we developed a mechanistic physiologically based pharmacokinetic model of the heart. The models were described with literature-derived parameters and written in R, v.3.4.0. Five parameters were estimated. The model was fitted to amitriptyline and nortriptyline concentrations after an intravenous infusion of amitriptyline. The cardiac model consisted of 5 compartments representing the pericardial fluid, heart extracellular water, and epicardial intracellular, midmyocardial intracellular, and endocardial intracellular fluids. Drug cardiac metabolism, passive diffusion, active efflux, and uptake were included in the model as mechanisms involved in the drug disposition within the heart. The model accounted for inter-individual variability. The estimates of optimized parameters were within physiological ranges. The model performance was verified by simulating 5 clinical studies of amitriptyline intravenous infusion, and the simulated pharmacokinetic profiles agreed with clinical data. The results support the model feasibility. The proposed structure can be tested with the goal of improving the patient-specific model-based cardiac safety assessment and offers a framework for predicting cardiac concentrations of various xenobiotics.

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.

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.

Empirical search for factors affecting mean particle size of PLGA microspheres containing macromolecular drugs

BACKGROUND AND OBJECTIVES

Poly(lactic-co-glycolic acid) (PLGA) has become one of the most promising in design, development, and optimization for medical applications polymers. PLGA-based multiparticulate dosage forms are usually prepared as microspheres where the size is from 5 to 100 µm, depending on the route of administration. The main objectives of the study were to develop a predictive model of mean volumetric particle size and on its basis extract knowledge of PLGA containing proteins forming behaviour.

METHODS

In the present study, a model for the prediction of mean volumetric particle size developed by an rgp package of R environment is presented. Other tools like fscaret, monmlp, fugeR, MARS, SVM, kNNreg, Cubist, randomForest and piecewise linear regression are also applied during the data mining procedure.

RESULTS

The feature selection provided by the fscaret package reduced the original input vector from a total of 295 input variables to 10, 16 and 19. The developed models had good predictive ability, which was confirmed by a normalized root-mean-square error (NRMSE) of 6.8 to 11.1% in 10-fold cross validation training procedure. Moreover, the best models were validated using external experimental data. The superior predictiveness had a model obtained by rgp in the form of a classical equation with a normalized root-mean-squared error (NRMSE) of 6.1%.

CONCLUSIONS

A new approach is proposed for computational modelling of the mean particle size of PLGA microspheres and rules extraction from tree-based models. The feature selection leads to revealing chemical descriptor variables which are important in predicting the size of PLGA microspheres. In order to achieve better understanding in the relationships between particle size and formulation characteristics, the surface analysis method and rules extraction procedures were applied.

Computational Intelligence Modeling of the Macromolecules Release from PLGA Microspheres-Focus on Feature Selection

Poly-lactide-co-glycolide (PLGA) is a copolymer of lactic and glycolic acid. Drug release from PLGA microspheres depends not only on polymer properties but also on drug type, particle size, morphology of microspheres, release conditions, etc. Selecting a subset of relevant properties for PLGA is a challenging machine learning task as there are over three hundred features to consider. In this work, we formulate the selection of critical attributes for PLGA as a multiobjective optimization problem with the aim of minimizing the error of predicting the dissolution profile while reducing the number of attributes selected. Four bio-inspired optimization algorithms: antlion optimization, binary version of antlion optimization, grey wolf optimization, and social spider optimization are used to select the optimal feature set for predicting the dissolution profile of PLGA. Besides these, LASSO algorithm is also used for comparisons. Selection of crucial variables is performed under the assumption that both predictability and model simplicity are of equal importance to the final result. During the feature selection process, a set of input variables is employed to find minimum generalization error across different predictive models and their settings/architectures. The methodology is evaluated using predictive modeling for which various tools are chosen, such as Cubist, random forests, artificial neural networks (monotonic MLP, deep learning MLP), multivariate adaptive regression splines, classification and regression tree, and hybrid systems of fuzzy logic and evolutionary computations (fugeR). The experimental results are compared with the results reported by Szlȩk. We obtain a normalized root mean square error (NRMSE) of 15.97% versus 15.4%, and the number of selected input features is smaller, nine versus eleven.

The Relationship Between the Evolution of an Internal Structure and Drug Dissolution from Controlled-Release Matrix Tablets

In the last decade, imaging has been introduced as a supplementary method to the dissolution tests, but a direct relationship of dissolution and imaging data has been almost completely overlooked. The purpose of this study was to assess the feasibility of relating magnetic resonance imaging (MRI) and dissolution data to elucidate dissolution profile features (i.e., kinetics, kinetics changes, and variability). Commercial, hydroxypropylmethyl cellulose-based quetiapine fumarate controlled-release matrix tablets were studied using the following two methods: (i) MRI inside the USP4 apparatus with subsequent machine learning-based image segmentation and (ii) dissolution testing with piecewise dissolution modeling. Obtained data were analyzed together using statistical data processing methods, including multiple linear regression. As a result, in this case, zeroth order release was found to be a consequence of internal structure evolution (interplay between region's areas-e.g., linear relationship between interface and core), which eventually resulted in core disappearance. Dry core disappearance had an impact on (i) changes in dissolution kinetics (from zeroth order to nonlinear) and (ii) an increase in variability of drug dissolution results. It can be concluded that it is feasible to parameterize changes in micro/meso morphology of hydrated, controlled release, swellable matrices using MRI to establish a causal relationship between the changes in morphology and drug dissolution. Presented results open new perspectives in practical application of combined MRI/dissolution to controlled-release drug products.

Transparent computational intelligence models for pharmaceutical tableting process

Purpose

Pharmaceutical industry is tightly regulated owing to health concerns. Over the years, the use of computational intelligence (CI) tools has increased in pharmaceutical research and development, manufacturing, and quality control. Quality characteristics of tablets like tensile strength are important indicators of expected tablet performance. Predictive, yet transparent, CI models which can be analysed for insights into the formulation and development process.

Methods

This work uses data from a galenical tableting study and computational intelligence methods like decision trees, random forests, fuzzy systems, artificial neural networks, and symbolic regression to establish models for the outcome of tensile strength. Data was divided in training and test fold according to ten fold cross validation scheme and RMSE was used as an evaluation metric. 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.

Results

CI models for tensile strength of tablets based on the formulation design and process parameters have been established. Best models exhibit normalized RMSE of 7 %. Rules from fuzzy systems and random forests are shown to increase transparency of CI models. A mathematical formula generated by symbolic regression is presented as a transparent model.

Conclusions

CI models explain the variation of tensile strength according to formulation and manufacturing process characteristics. CI models can be further analyzed to extract actionable knowledge making the artificial learning process more transparent and acceptable for use in pharmaceutical quality and safety domains.

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.

Roughness-controlled self-assembly of mannitol/LB agar microparticles by polymorphic transformation for pulmonary drug delivery

Novel roughness-controlled mannitol/LB Agar microparticles were synthesized by polymorphic transformation and self-assembly method using hexane as the polymorphic transformation reagent and spray-dried mannitol/LB Agar microparticles as the starting material. As-prepared microparticles were characterized by Fourier transform infrared spectra (FTIR), X-ray diffraction spectra (XRD), differential scanning calorimetry (DSC), scanning electron microscopy (SEM), thermal gravimetric analysis (TGA), and Andersen Cascade Impactor (ACI). The XRD and DSC results indicate that after immersing spray-dried mannitol/LB Agar microparticles in hexane, β-mannitol was completely transformed to α-mannitol in 1 h, and all the δ-mannitol was transformed to α form after 14 days. SEM shows that during the transformation the nanobelts on the spray-dried mannitol/LB Agar microparticles become more dispersed and the contour of the individual nanobelts becomes more noticeable. Afterward, the nanobelts self-assemble to nanorods and result in rod-covered mannitol/LB Agar microparticles. FTIR indicates new hydrogen bonds were formed among mannitol, LB Agar, and hexane. SEM images coupled with image analysis software reveal that different surface morphology of the microparticles have different drug adhesion mechanisms. Comparison of ACI results and image analysis of SEM images shows that an increase in the particle surface roughness can increase the fine particle fractions (FPFs) using the rod-covered mannitol microparticles as drug carriers. Transformed microparticles show higher FPFs than commercially available lactose carriers. An FPF of 28.6 ± 2.4% was achieved by microparticles transformed from spray-dried microparticles using 2% mannitol(w/v)/LB Agar as feed solution. It is comparable to the highest FPF reported in the literature using lactose and spray-dried mannitol as carriers.

Computer-based prediction of the drug proarrhythmic effect: problems, issues, known and suspected challenges

It is likely that computer modelling and simulations will become an element of comprehensive cardiac safety testing. Their role would be primarily the integration and the interpretation of previously gathered data. There are still unanswered questions and issues which we list and describe below. They include sources of data used for the development of the models as well as data utilized as input information, which can come from the in vitro studies and the quantitative structure-activity relationship models. The pharmacokinetics of the drugs in question play a crucial role as their active concentration should be considered, yet the question remains where is the right place to assess it. The pharmacodynamic angle includes complications coming from multiple drugs (i.e. active metabolites) acting in parallel as well as the type of interaction with (potentially) multiple affected channels. Once established, the model and the methodology of its use should be further validated, optimistically against individual data reported at the clinical level as the physiological, anatomical, and genetic parameters play a crucial role in the drug-triggered arrhythmia induction. All the abovementioned issues should be at least considered and-hopefully-resolved, to properly utilize the mathematical models for a cardiac safety assessment.

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.

In vitro-in vivo extrapolation of drug-induced proarrhythmia predictions at the population level

Drug cardiotoxicity is a serious issue for patients, regulators, pharmaceutical companies and health service payers because they are all affected by its consequences. Despite the wide range of data they generate, existing approaches for cardiac safety testing might not be adequate and sufficiently cost-effective, probably as a result of the complexity of the problem. For this reason, translational tools (based on biophysically detailed, mathematical models) allowing for in vitro-in vivo extrapolation are gaining increasing interest. This current review describes approaches that can be used for cardiac safety assessment at the population level, by accounting for various sources of variability including kinetics of the compound of interest.

Generalized in vitro-in vivo relationship (IVIVR) model based on artificial neural networks

BACKGROUND

The aim of this study was to develop a generalized in vitro-in vivo relationship (IVIVR) model based on in vitro dissolution profiles together with quantitative and qualitative composition of dosage formulations as covariates. Such a model would be of substantial aid in the early stages of development of a pharmaceutical formulation, when no in vivo results are yet available and it is impossible to create a classical in vitro-in vivo correlation (IVIVC)/IVIVR.

METHODS

Chemoinformatics software was used to compute the molecular descriptors of drug substances (ie, active pharmaceutical ingredients) and excipients. The data were collected from the literature. Artificial neural networks were used as the modeling tool. The training process was carried out using the 10-fold cross-validation technique.

RESULTS

The database contained 93 formulations with 307 inputs initially, and was later limited to 28 in a course of sensitivity analysis. The four best models were introduced into the artificial neural network ensemble. Complete in vivo profiles were predicted accurately for 37.6% of the formulations.

CONCLUSION

It has been shown that artificial neural networks can be an effective predictive tool for constructing IVIVR in an integrated generalized model for various formulations. Because IVIVC/IVIVR is classically conducted for 2-4 formulations and with a single active pharmaceutical ingredient, the approach described here is unique in that it incorporates various active pharmaceutical ingredients and dosage forms into a single model. Thus, preliminary IVIVC/IVIVR can be available without in vivo data, which is impossible using current IVIVC/IVIVR procedures.

Predictive model for L-type channel inhibition: multichannel block in QT prolongation risk assessment

Evaluation of the proarrhythmic potential of an investigated compound is now an integral element of the safety profile required for the approval of new drugs. The human ether-à-go-go-related gene (hERG) channel blocking potency is regarded as a surrogate marker of the proarrhythmic risk at the early stages of the research and development process. However, there is no straight correlation between QT prolongation and TdP occurrence probability, and hERG inhibition potential can be an inadequate predictor of QT prolongation. The L-type calcium channel plays a pivotal role in cardiomyocytes' physiology. Thus the main aim of this study was to develop a predictive model for drug-triggered CaL channel inhibition and also the assessment of drug-multichannel interaction effects on the heart rate-corrected QT interval. The data set, consisting of 123 records describing in vitro experimental settings, measured IC₅₀ values and calculated physico-chemical properties for 72 various chemicals, was collected. The models were tested in a modified 10-fold cross-validation procedure. The generalization ability of the best model was as follows: root mean squared error (RMSE) = 1.10, normalized root mean squared error (NRMSE) = 16.09%. Out of the 10 most important variables, 5 described conditions of the in vitro experiments thus their description and experiment's conditions standardization might be the key to the models better performance. The simulations performed with the ToxComp system showed that the hERG block alone causes concentration-dependent QT prolongation, whereas when multichannel block is regarded, the effect could be reversed. For that reason, the multichannel interaction of tested compounds should be taken into consideration, in order to make the proarrhythmic risk assessment more reliable.

Virtual population generator for human cardiomyocytes parameters: in silico drug cardiotoxicity assessment

BACKGROUND

The anatomical and histological parameters of the human ventricle depend on many factors including age and sex. Myocyte volume and electric capacitance are significant physiological parameters of left ventricle cardiomyocyte mathematical models. They allow the assessment of inter-individual variability during in vitro-in vivo extrapolation of the drug cardiotoxic effect.

OBJECTIVE

The current research was carried out to analyze the relationship between age, human left ventricle cardiomyocyte volume, and electric capacitance in a healthy population.

METHODS

In order to collect data describing cardiomyocyte volume and membrane area, literature searches were performed. It was assumed that the cardiomyocyte volume (VOL) and area (AREA) distribution have non-negative support and are skewed to the right. A log-linear model with constant variance was used. A simulation study was run to assess the influence of physiological parameters on action potential duration.

RESULTS

The coefficient of determination for the proposed model R(2) = 0.95, that is, 95% of the variability observed in log cardiomyocyte volume can be explained by the estimated regression equation. To allow simple calculation and model performance validation, a simple Excel file was developed (Supplementary material).

CONCLUSIONS

To the best of our knowledge, there is no other model available, combining age, cardiomyocyte volume, and area. The main limitations of the proposed models result from the assumptions made at the data analysis stage. The limited amount of information available in the literature and the lack of differentiation between sexes results in one common equation. The developed model is a part of the computational system for drug cardiotoxicity assessment.

Slow delayed rectifying potassium current (IKs ) - analysis of the in vitro inhibition data and predictive model development

The excitable cell membranes contain ion channels that allow the ions passage through the specific pores via a passive process. Assessment of the inhibition of the IKr (hERG) current is considered to be the main target during the drug development process, although there are other ionic currents for which drug-triggered modification can either potentiate or mask hERG channel blockade. Information describing the results of in vitro studies investigating the chemical-IKs current interactions has been developed in the current study. Based on the publicly available data sources, 145 records were collected. The final list of publications consists of 64 positions and refers to 106 different molecules connected with IKs current inhibition, with at least one IC50 value measured. Ultimately, 98 of the IC50 values expressed as absolute values were gathered. For 36 records the IC50 was expressed as a relative value. For the 11 remaining records, the inhibition was not clearly expressed. Based on the collected data the predictive models for the IC50 estimation were developed with the use of various algorithms. The extended Quantitative Structure-Activity Relationships (QSAR) methodology was applied and the in vitro research settings were included as independent variables, apart from the physico-chemical descriptors calculated with the use of the Marvin Calculator Plugins. The root mean squared error and normalized root mean squared error values for the best model (an expert system based on two independent artificial neural networks) were 0.86 and 14.04%, respectively. The model was further built into the ToxComp system, the ToxIVIVE tool specialized for cardiotoxicity assessment of drugs.

The open-access dataset for insilico cardiotoxicity prediction system

Drug cardiotoxicity is one of the main reasons of fatal drug related problem events and the subsequent withdrawals. Therefore, its early assessment is a crucial element of the drug development process. For the drug driven hERG inhibition assessment, which is assumed to be the main reason for toxicity, in vitro techniques are used. Gold standards are based on the Patch Clamp method with the use of various cell models but due to its low throughput, insilico models have become more appreciated. To develop a reliable empirical QSAR model, wide dataset containing a variety of cases has to be available. In this article, a freely available for download, set of data is described. It is based on literature peer-reviewed reports and contains hERG inhibition information expressed as IC50 value for 263 molecules described in 642 records. All studies were done with the use of three cell models (XO, CHO, HEK) and other elements describe the electrophysiological settings of the in vitro study. The above mentioned set was used for the successful development of the predictive models.

Novel application of MRI technique combined with flow-through cell dissolution apparatus as supportive discriminatory test for evaluation of controlled release formulations

Dissolution studies cannot distinguish phenomena occurring inside the dosage forms when studying formulation with similar dissolution profiles-such formulations can behave differently when considering their physical changes. The application of flow-through dissolution apparatus integrated with magnetic resonance imaging (MRI) system for discriminative evaluation of controlled release dosage forms with similar dissolution profiles was presented. Hydrodynamically balanced systems (HBS) containing L: -dopa and various grades hydroxypropyl methylcelluloses were prepared. The dissolution studies of L: -dopa were performed at high field (4.7 T) MR system with MR-compatible flow-through cell. MRI was done with 0.14 x 0.14 x 1-mm spatial resolution and temporal resolution of 10 min to record changes of HBS parameters during dissolution in 0.1 M HCl. Structural and geometrical changes were evaluated using the following parameters: total area of HBS cross-section, its Feret's diameter, perimeter and circularity, area of hydrogel layer, and "dry core" area. While the dissolution profiles of L: -dopa were similar, the image analysis revealed differences in the structural and geometrical changes of the HBS. The mechanism of drug release from polymeric matrices is a result of synergy of several different phenomena occurring during dissolution and may differ between formulations, yet giving similar dissolution profiles. A multivariate analysis was performed to create a model taking into account dissolution data, data from MRI, information about chemical structure, and polymer viscosity. It provided a single model for all the formulations which was confirmed to be competent. The presented method has merit as a potential Process Analytical Technology tool.

The usefulness of artificial neural networks in the evaluation of pulmonary efficiency and antioxidant capacity of welders

OBJECTIVE

The aim of the study was to determine if artificial neural networks (ANNs) may be useful to analyse a complex and large set of data derived from smoking welders for the purpose of finding relationships between parameters describing respiratory system efficiency and antioxidant defence.

METHODS

A group of 94 welders employed in a big metallurgic enterprise in Krakow, Poland (men only, aged 29-57 years, all active smokers) occupationally exposed to O(3) and NO(x), were the subjects of this study. They underwent biochemical measurements including total antioxidant status (TAS) and the anti-oxidative defence enzymes superoxide dismutase (SOD) and catalase (CT); biominerals: Fe, Cu, Zn, Mg in blood serum and in hair; the concentrations of albumin, bilirubin, uric acid in blood. The determination of respiratory efficiency was based on a "flow-volume" curve and spirometry. The dependant variables for ANNs were: TAS, SOD, CT. Thirty-one subjects with normal values of all spirometric parameters were selected for the final analysis.

RESULTS

Statistically valid relationship between TAS and albumin, Zn and Cu in blood and the two pulmonary parameters forced expiratory volume after 1s (FEV(1)) and middle expiratory flow of 25-75% of vital capacity (MEF(25/75)) were found. Zn concentration almost linearly influenced TAS. For Cu a sigmoid curve was obtained. For albumin concentration as well as for FEV(1) a two-stage curve was observed.

CONCLUSIONS

ANNs are useful for the reduction of dimensionality and are suited to analyse a complex and relatively large set of parameters when it is unknown which of these are related. ANNs were useful for finding the relationship between the antioxidant defence and the respiratory system capacity in welders who smoke.

Micronutrients and macronutrients and parameters of antioxidative ability in saliva of women: inhabitants of Krakow (Poland) in the course of uncomplicated singleton pregnancy

There were two aims of the present study: (1) to evaluate changes of superoxide dismutase (SOD) activity and total antioxidative status measured as the ferric reducing ability of plasma (FRAP) concentration in saliva of pregnant women during the first, second, and third trimesters of singleton uncomplicated pregnancy and (2) to assess possible relations among SOD, FRAP, and intake of macronutrients and micronutrients in daily nutritional rations (DNRs) during pregnancy. Forty pregnant women aged 27.1+/-5.4 yr (examined group) and 40 healthy women (the control group) were recruited for this study. The relationship between FRAP and SOD in saliva and the intake of macronutrients (proteins, carbohydrates, total fat, saturated fatty acid, monounsaturated fatty acids, polyunsaturated fatty acids), minerals (Na, K, Ca, P, Mg, Fe, Zn, Cu, Mn), and vitamins (A, C, E, B1, B2, B6, PP) in DNR was evaluated by clustering analysis with Ward's grouping method. During pregnancy, FRAP and SOD values were lower than in the controls, but only for FRAP were the differences statistically significant (p < 0.001). For the whole pregnancy period, cluster analysis identified two major clusters for which the differentiating variables were SOD and retinoids intake, but different patterns for each trimester of pregnancy were revealed. The following were concluded: (1) FRAP values were the lowest in the second trimester. It suggests that in this trimester of uncomplicated pregnancy, women might be not fully adapted to increased demands for antioxidative mechanisms. (2) Cluster analysis showed that there were no statistical relationships between the intake of micronutrients and macronutrients in DNRs and the SOD or FRAP level in the saliva of pregnant women.

Artificial neural networks in the modeling of drugs release profiles from hydrodynamically balanced systems

Artificial neural networks (ANNs) were used as modeling tools for prediction of various drugs release patterns from hydrodynamically balanced systems (HBS) composed with Metholose 90SH (hydroxypropylmethylcellulose--HPMC). The objective was to provide predictive and data-mining models of analyzed problem. It was found that ANNs are capable to accurately predict release patterns of different drugs from HBS based on the description of the formulation as well as chemical structure of the drug. Overall generalization error RMSE was 8.7 and after inclusion of pilot study in learning dataset it decreased to ca. 4.5. Sensitivity analysis of ANNs was applied to reduce native input vector from 77 to 7 inputs in order to improve the performance of predictive models. Simultaneously, it revealed crucial variables governing release of drugs from HBS.

Artificial neural network in pharmacoeconomics

Pharmacoeconomics research identifies, measures, and compares the costs (resources consumed) and consequences of medical products and services, where at least one of the compared alternatives is pharmacotherapy. Pharmacoeconomics has been designed to enable the decision maker to identify the preferred choice among existing alternatives. The decisions are often important for the patients' lives on the one hand, and for payers on the other (where the payer is understood as the institution responsible for financial resources allocation). One of the most commonly used types of pharmacoeconomic analysis is cost-effectiveness analysis. Two different alternatives can be compared using cost-effectiveness analysis if only their medical and clinical consequences could be measured in similar units (clinical or physical parameters). The aim of the project is to use an artificial neural network (ANN) for medical effect prediction, which could help in the extrapolation of pharmacoeconomic analysis' results. To depict neural data analysis tools, a database containing 100 non-small cell lung cancer (NSCLC) patients in non-operative IIIB and IV stage has been used. Each patient was described using 30 factors (i.e. sex, age, anticancer drugs dosage) and, as an output value, the expected survival time was established. The role of the ANN system was to predict the patient's survival time based on the above mentioned information. Binary values were tested as outcomes. Positive values (coded as 1) meant that patient survival time would be equal to or longer than 35 weeks. Negative values (coded as 0) meant that the patient survival time would be shorter than 35 weeks. Binary values were obtained using a threshold, which based on the mean survival time of patients derived from literature. Back-propagation as well as fuzzy-logic neural networks were applied. A 10-fold cross validation method was used to obtain the appropriate models. Final results were compared with the generic, logistic regression-based model. The best prediction score of the ANN model was 82%; higher than logistic regression prediction rate.

Artificial neural networks as an engine of Internet based hypertension prediction tool

Hypertension is the most common cause of death. Therefore it is recognized as a major civilization disease next to diabetes, hyperuricemia, asthma etc. The objective was to use artificial neural networks (ANNs) to handle demographic data and to produce system of hypertension risk prediction. Database used in the development of hypertension risk model was obtained from CDC (BRFSS--Behavioral Risk Factor Surveillance System). Screening for optimal ANN architecture was performed among various backpropagation and fuzzy neural networks with use of 10-fold cross-validation scheme. Single ANNs as well as experts committees were tested. Best results were found to be around 75%--expressed as total classification rate. Java applet was designed to be the interface between ANN system and end user. Spreadsheet form was chosen to facilitate navigation and used by healthcare non-specialists. Free of charge Internet publication is expected soon at the address [url: see text].

Influence of new gamma-aminobutyric acid amide derivatives and its phthalimide precursors on the central nervous system activity in mice

The present study investigated the influence of BM-78, BM-121 (gamma-aminobutyric acid amide derivatives) and BM-42, BM-43 (phthalimide precursors of BM-78 and BM-121) on the spontaneous locomotor activity and on the picrotoxin-induced seizures. Results of pharmacological in vivo examination of the effects of new gamma-aminobutyric acid amide derivatives and its phthalimide precursors (compounds BM-78, BM-121, BM-42, BM-43), presented in this paper showed that all the compounds had different but clear influence on CNS in mice.