Machine Learning-Enabled NIR Spectroscopy. Part 3: Hyperparameter by Design (HyD) Based ANN-MLP Optimization, Model Generalizability, and Model Transferability.

Data variations, library changes, and poorly tuned hyperparameters can cause failures in data-driven modelling. In such scenarios, model drift, a gradual shift in model performance, can lead to inaccurate predictions. Monitoring and mitigating drift are vital to maintain model effectiveness. USFDA and ICH regulate pharmaceutical variation with scientific risk-based approaches. In this study, the hyperparameter optimization for the Artificial Neural Network Multilayer Perceptron (ANN-MLP) was investigated using open-source data. The design of experiments (DoE) approach in combination with target drift prediction and statistical process control (SPC) was employed to achieve this objective. First, pre-screening and optimization DoEs were conducted on lab-scale data, serving as internal validation data, to identify the design space and control space. The regression performance metrics were carefully monitored to ensure the right set of hyperparameters was selected, optimizing the modelling time and storage requirements. Before extending the analysis to external validation data, a drift analysis on the target variable was performed. This aimed to determine if the external data fell within the studied range or required retraining of the model. Although a drift was observed, the external data remained well within the range of the internal validation data. Subsequently, trend analysis and process monitoring for the mean absolute error of the active content were conducted. The combined use of DoE, drift analysis, and SPC enabled trend analysis, ensuring that both current and external validation data met acceptance criteria. Out-of-specification and process control limits were determined, providing valuable insights into the model's performance and overall reliability. This comprehensive approach allowed for robust hyperparameter optimization and effective management of model lifecycle, crucial in achieving accurate and dependable predictions in various real-world applications.

Influence of Crystal Disorder on the Forced Oxidative Degradation of Vortioxetine HBr.

The present study investigates the impact of the solid-state disorder of vortioxetine hydrobromide (HBr) on oxidative degradation under accelerated conditions. A range of solid-state disorders was generated via cryogenic ball milling. The solid-state properties were evaluated by calorimetry, infrared-, and Raman spectroscopies. While salt disproportionation occurred upon milling, no chemical degradation occurred by milling. The amorphous fraction remained physically intact under ambient storage conditions. Subsequently, samples with representative disordered fractions were mixed with a solid oxidative stressor (PVP-H2O2 complex) and were compressed to compacts. The compacts were exposed to 40°C/75% RH for up to 6 h. The sample was periodically withdrawn and analyzed for the physical transformations and degradation. Two oxidative degradation products (DPs) were found to be formed, for which dissimilar relations to the degree of disorder and kinetics of formation were observed. The degradation rate of the major DP formation obtained by fitting the exponential model to the experimental data was found to increase up to a certain degree of disorder and decrease with a further increase in the disordered fraction. In contrast, the minor DP formation kinetics was found to increase monotonically with the increase in the disorder content. For the similar crystallinity level, the degradation trend (rate and extent) differed between the single-phase disorder generated by milling and physically mixed two-phase systems. Overall, the study demonstrates the importance of evaluating the physical and chemical (in)stabilities of the disordered solid state of a salt form of a drug substance, generated through mechano-activation.

Investigation of the Impact of Saccharides on the Relative Activity of Trypsin and Catalase after Droplet and Spray Drying.

While using saccharides as stabilizers for therapeutic protein drying is common, the mechanisms underlying the stabilization during drying remain largely unexplored. Herein, we investigated the effect of different saccharides, trehalose dihydrate (TD), dextran (DEX), and hydroxypropyl ß-cyclodextrins (low substitution-HP and high substitution-HPB), on the relative activities of the enzymes trypsin and catalase during miniaturized drying (MD) or spray drying (SD). For trypsin, the presence of saccharides, especially HP, was beneficial, as it significantly improved the enzyme activity following MD. The HPB preserved trypsin's activity during MD and SD. Adding saccharides during MD did not show a notable improvement in catalase activities. Increasing TD was beneficial during the SD of catalase, as indicated by significantly increased activity. Molecular docking and molecular dynamics simulations oftrypsin with HP or HPB revealed the influence of their substitution on the binding affinity for the enzyme. A higher affinity of HP to bind trypsin and itself was observed during simulations. Experimentally, activity reduction was mainly observed during MD, attributable to the higher droplet temperature during MD than during SD. The activities from the experiments and aggregation propensity from molecular modeling helped elucidate the impact of the size of protein and saccharides on preserving the activity during drying.

pH stimuli-responsive hydrogels from non-cellulosic biopolymers for drug delivery.

Over the past several decades, there has been significant growth in the design and development of more efficient and advanced biomaterials based on non-cellulosic biological macromolecules. In this context, hydrogels based on stimuli-responsive non-cellulosic biological macromolecules have garnered significant attention because of their intrinsic physicochemical properties, biological characteristics, and sustainability. Due to their capacity to adapt to physiological pHs with rapid and reversible changes, several researchers have investigated pH-responsive-based non-cellulosic polymers from various materials. pH-responsive hydrogels release therapeutic substances in response to pH changes, providing tailored administration, fewer side effects, and improved treatment efficacy while reducing tissue damage. Because of these qualities, they have been shown to be useful in a wide variety of applications, including the administration of chemotherapeutic drugs, biological material, and natural components. The pH-sensitive biopolymers that are utilized most frequently include chitosan, alginate, hyaluronic acid, guar gum, and dextran. In this review article, the emphasis is placed on pH stimuli-responsive materials that are based on biological macromolecules for the purposes of drug administration.

Pancreatic lipase digestion: The forgotten barrier in oral administration of lipid-based delivery systems?

This review highlights the importance of controlling the digestion process of orally administered lipid-based delivery systems (LBDS) and their performance. Oral LBDS are prone to digestion via pancreatic lipase in the small intestine. Rapid or uncontrolled digestion may cause the loss of delivery system integrity, its structural changes, reduced solubilization capacity and physical stability issues. All these events can lead to uncontrolled drug release from the digested LBDS into the gastrointestinal environment, exposing the incorporated drug to precipitation or degradation by luminal proteases. To prevent this, the digestion rate of orally administered LBDS can be estimated by appropriate choice of the formulation type, excipient combinations and their ratios. In addition, in vitro digestion models like pH-stat are useful tools to evaluate the formulation digestion rate. Controlling digestion can be achieved by conventional lipase inhibitors like orlistat, sterically hindering of lipase adsorption on the delivery system surface with polyethylene glycol (PEG) chains, lipase desorption or saturation of the interface with surfactants as well as formulating LBDS with ester-free excipients. Recent in vivo studies demonstrated that digestion inhibition lead to altered pharmacokinetic profiles, where Cmax and Tmax were reduced in spite of same AUC compared to control or even improved oral bioavailability.

Neutralizing Antibody Validation Testing and Reporting Harmonization.

Evolving immunogenicity assay performance expectations and a lack of harmonized neutralizing antibody validation testing and reporting tools have resulted in significant time spent by health authorities and sponsors on resolving filing queries. A team of experts within the American Association of Pharmaceutical Scientists' Therapeutic Product Immunogenicity Community across industry and the Food and Drug Administration addressed challenges unique to cell-based and non-cell-based neutralizing antibody assays. Harmonization of validation expectations and data reporting will facilitate filings to health authorities and are described in this manuscript. This team provides validation testing and reporting strategies and tools for the following assessments: (1) format selection; (2) cut point; (3) assay acceptance criteria; (4) control precision; (5) sensitivity including positive control selection and performance tracking; (6) negative control selection; (7) selectivity/specificity including matrix interference, hemolysis, lipemia, bilirubin, concomitant medications, and structurally similar analytes; (8) drug tolerance; (9) target tolerance; (10) sample stability; and (11) assay robustness.

On the influence of raw material attributes on process behaviour and product quality in a continuous WET granulation tableting line.

Continuous manufacturing of oral solids is a complex process in which critical material attributes (CMAs), formulation and critical process parameters (CPPs) play a fundamental role. However, assessing their effect on the intermediate and final product's critical quality attributes (CQAs) remains challenging. The aim of this study was to tackle this shortcoming by evaluating the influence of raw material properties and formulation composition on the processability and quality of granules and tablets on a continuous manufacturing line. Powder-to-tablet manufacturing was performed using four formulations in various process settings. Pre-blends of different drug loadings (2.5 % w/w and 25% w/w) and two BCS classes (Class I and II) were continuously processed on an integrated process line ConsiGmaTM 25, including twin screw wet granulation, fluid bed drying, milling, sieving, in-line lubrication and tableting. The liquid-to-solid ratio and the granule drying time were varied to process granules under nominal, dry and wet conditions. It was shown that the BCS class and the drug dosage influenced the processability. Intermediate quality attributes, such as the loss on drying and the particle size distribution, directly correlated with the raw material's properties and process parameters. Process settings had a profound impact on the tablet's hardness, disintegration time, wettability and porosity.

Development of a Digital Interface for Personalized Dosing in Renal Impaired Patients: A Case-Study Using the ACE-Inhibitor Benazepril.

BACKGROUND: Personalized dosing regimens have great potential to improve the standard level of care from "one-fits-all" to the "right dose, to the right patient at the right time". OBJECTIVES: Development of a digital interface that can inform healthcare professionals on the dosing of an ACE inhibitor on an individual basis. METHODS: A physiologically based pharmacokinetic (PBPK) model and a one-compartment model were implemented for the prodrug benazepril and its metabolite benazeprilat, respectively. In sequence, to capture inter-individual differences the models were extended to a population based one (PopPBPK). RESULTS: Both models predicted the pharmacokinetic data in the observed ranges. Application of the models help identify the factors influencing drug concentrations in the body and to find subgroups of patients, in which a dose adjustment is recommended, or a higher degree of caution is required. CONCLUSION: The use of the models via a practical user interface can help inform clinical decisions and design optimal dosing based on the individual anthropometric characteristics and stage of renal impairment.

Applying Material Science Principles to Chemical Stability: Modelling Solid State Autoxidation in Mifepristone Containing Different Degrees of Crystal Disorder.

Ball-milling and harsh manufacturing processes often generate crystal disorder which have practical implications on the physical and chemical stabilities of solid drugs during subsequent storage, transport, and handling. The impact of the physical state of solid drugs, containing different degrees/levels of crystal disorder, on their autoxidative stability under storage has not been widely investigated. This study investigates the impact of differing degrees of crystal disorder on the autoxidation of Mifepristone (MFP) to develop a predictive (semi-empirical) stability model. Crystalline MFP was subjected to different durations of ambient ball milling, and the resulting disorder/ amorphous content was quantified using a partial least square (PLS) regression model based on Raman spectroscopy data. Samples of MFP milled to generate varying levels of disorder were subjected to a range of (accelerated) stability conditions, and periodically sampled to examine their recrystallization and degradation extents. Crystallinity was monitored by Raman spectroscopy, and the degradation was evaluated by liquid chromatography. The analyses of milled samples demonstrated a competition between recrystallization and degradation via autoxidation of MFP, to different extents depending on stability conditions/exposure time. The degradation kinetics were analyzed by accounting for the preceding amorphous content, and fitted with a diffusion model. An extended Arrhenius equation was used to predict the degradation of stored samples under long-term (25°C/60% RH) and accelerated (40°C/75% RH, 50°C/75% RH) stability conditions. This study highlights the utility of such a predictive stability model for identifying the autoxidative instability in non-crystalline/partially crystalline MFP, owing to the degradation of the amorphous phases. This study is particularly useful for identifying drug-product instability by leveraging the concept of material sciences.

Impact of Different Saccharides on the In-Process Stability of a Protein Drug During Evaporative Drying: From Sessile Droplet Drying to Lab-Scale Spray Drying.

OBJECTIVES: Solid biopharmaceutical products can circumvent lower temperature storage and transport and increase remote access with lower carbon emissions and energy consumption. Saccharides are known stabilizers in a solid protein produced via lyophilization and spray drying (SD). Thus, it is essential to understand the interactions between saccharides and proteins and the stabilization mechanism. METHODS: A miniaturized single droplet drying (MD) method was developed to understand how different saccharides stabilize proteins during drying. We applied our MD to different aqueous saccharide-protein systems and transferred our findings to SD. RESULTS: The poly- and oligosaccharides tend to destabilize the protein during drying. The oligosaccharide, Hydroxypropyl ß-cyclodextrin (HPßCD) shows high aggregation at a high saccharide-to-protein molar ratio (S/P ratio) during MD, and the finding is supported by nanoDSF results. The polysaccharide, Dextran (DEX) leads to larger particles, whereas HPBCD leads to smaller particles. Furthermore, DEX is not able to stabilize the protein at higher S/P ratios either. In contrast, the disaccharide Trehalose Dihydrate (TD) does not increase or induce protein aggregation during the drying of the formulation. It can preserve the protein's secondary structure during drying, already at low concentrations. CONCLUSION: During the drying of S/P formulations containing the saccharides TD and DEX, the MD approach could anticipate the in-process (in) stability of protein X at laboratory-scale SD. In contrast, for the systems with HPßCD, the results obtained by SD were contradictory to MD. This underlines that depending on the drying operation, careful consideration needs to be applied to the selection of saccharides and their ratios.

Screening Autoxidation Propensities of Drugs in the Solid-State Using PVP and in the Solution State Using N-Methyl Pyrrolidone.

Oxidative degradation of drugs is one of the major routes of drug substance and drug product instability. Among the diverse routes of oxidation, autoxidation is considered to be challenging to predict and control, potentially due to the multi-step mechanism involving free radicals. C-H bond dissociation energy (C-H BDE) is evidenced to be a calculated descriptor shown to predict drug autoxidation. While computational predictions for the autoxidation propensity of drugs are both swift and possible, no literature to date has highlighted the relationship between the computed C-H BDE and the experimentally-derived autoxidation propensities of solid drugs. The objective of this study is to investigate this missing relationship. The present work is an extension to the previously reported novel autoxidation approach that involves subjecting a physical mixture of pre-milled polyvinyl pyrrolidone (PVP) K-60 and a crystalline drug under high temperature and pressurized oxygen setup. The drug degradation was measured using chromatographic methods. An improved trend between the extent of solid autoxidation and C-H BDE could be observed after normalizing the effective surface area of drugs in the crystalline state, pointing to a positive relationship. Additional studies were conducted by dissolving the drug in N-methyl pyrrolidone (NMP) and exposing the solution under a pressurized oxygen setup at diverse elevated temperatures. Chromatographic results of these samples indicated a similarity in the formed degradation products to the solid-state experiments pointing to the utility of NMP, a PVP monomer surrogate, as a stressing agent for faster and relevant autoxidation screening of drugs in formulations.

Assessment of Tribo-charging and Continuous Feeding Performance of Direct Compression Grades of Isomalt and Mannitol Powders.

Tribo-charging is often a root cause of mass flow deviations and powder adhesion during continuous feeding. Thus, it may critically impact product quality. In this study, we characterized the volumetric (split- and pre-blend) feeding behavior and process-induced charge of two direct compression grades of polyols, galenIQ™ 721 (G721) for isomalt and PEARLITOL® 200SD (P200SD) for mannitol, under different processing conditions. The feeding mass flow range and variability, hopper end fill level, and powder adhesion were profiled. The feeding-induced tribo-charging was measured using a Faraday cup. Both materials were comprehensively characterized for relevant powder properties, and their tribo-charging was investigated for its dependence on particle size and relative humidity. During split-feeding experiments, G721 showed a comparable feeding performance to P200SD with lower tribo-charging and adhesion to the screw outlet of the feeder. Depending on the processing condition, the charge density of G721 ranged from -0.01 up to -0.39 nC/g, and for P200SD from -3.19 up to -5.99 nC/g. Rather than differences in the particle size distribution of the two materials, their distinct surface and structural characteristics were found as the main factors affecting their tribo-charging. The good feeding performance of both polyol grades was also maintained during pre-blend feeding, where reduced tribo-charging and adhesion propensity was observed for P200SD (decreasing from -5.27 to -0.17 nC/g under the same feeding settings). Here, it is proposed that the mitigation of tribo-charging occurs due to a particle size-driven mechanism.

Data-Driven Prediction of the Formation of Co-Amorphous Systems.

Co-amorphous systems (COAMS) have raised increasing interest in the pharmaceutical industry, since they combine the increased solubility and/or faster dissolution of amorphous forms with the stability of crystalline forms. However, the choice of the co-former is critical for the formation of a COAMS. While some models exist to predict the potential formation of COAMS, they often focus on a limited group of compounds. Here, four classes of combinations of an active pharmaceutical ingredient (API) with (1) another API, (2) an amino acid, (3) an organic acid, or (4) another substance were considered. A model using gradient boosting methods was developed to predict the successful formation of COAMS for all four classes. The model was tested on data not seen during training and predicted 15 out of 19 examples correctly. In addition, the model was used to screen for new COAMS in binary systems of two APIs for inhalation therapy, as diseases such as tuberculosis, asthma, and COPD usually require complex multidrug-therapy. Three of these new API-API combinations were selected for experimental testing and co-processed via milling. The experiments confirmed the predictions of the model in all three cases. This data-driven model will facilitate and expedite the screening phase for new binary COAMS.

Theoretical and Experimental Investigation of Autoxidation Propensity of Selected Drugs in Solution State.

The C-H bond dissociation energy (BDE) of drug molecules is often used to estimate their relative propensities to undergo autoxidation. BDE calculations based on electronic structures provide a convenient means to estimate the risk for a given compound to degrade via autoxidation. This study aimed to verify the utility of calculated C-H BDEs of a range of drug molecules in predicting their autoxidation propensities, in the solution state. For the autoxidation study, 2,2'-azobis (2-methylpropionitrile) was employed as the solution state stressor, and the experimental reaction rate constants were determined employing ultraperformance liquid chromatographic (UPLC) methods. Reaction rates in the solution state were compared to the calculated C-H BDE values of the respective compounds. The results indicated a poor correlation for compounds in the solution state, and their relative stabilities could not be explained with C-H BDE. On the other hand, a favorable relationship was observed between the relative extent of ionization and the autoxidation rates of the selected compounds. In the solution state, factors such as the type and extent of drug ionization, degree and type of solvation have been shown to contribute to differences in reactivity. By applying the computational method involving the effect of H-atom abstraction and potential ionization sites in the molecule, the calculated C-H BDE should relate better to the experimental autoxidation rates.

Role of Crystal Disorder and Mechanoactivation in Solid-State Stability of Pharmaceuticals.

Common energy-intensive processes applied in oral solid dosage development, such as milling, sieving, blending, compaction, etc. generate particles with surface and bulk crystal disorder. An intriguing aspect of the generated crystal disorder is its evolution and repercussion on the physical- and chemical stabilities of drugs. In this review, we firstly examine the existing literature on crystal disorder and its implications on solid-state stability of pharmaceuticals. Secondly, we discuss the key aspects related to the generation and evolution of crystal disorder, dynamics of the disordered/amorphous phase, analytical techniques to measure/quantify them, and approaches to model the disordering propensity from first principles. The main objective of this compilation is to provide special impetus to predict or model the chemical degradation(s) resulting from processing-induced manifestation in bulk solid manufacturing. Finally, a generic workflow is proposed that can be useful to investigate the relevance of crystal disorder on the degradation of pharmaceuticals during stability studies. The present review will cater to the requirements for developing physically- and chemically stable drugs, thereby enabling early and rational decision-making during candidate screening and in assessing degradation risks associated with formulations and processing.

Influence of PLGA End Groups on the Release Profile of Dexamethasone from Ocular Implants.

The present study deals with the development of dexamethasone (DM)-loaded implants using ester end-capped Resomer RG 502 poly(lactic acid-co-glycolic acid) (PLGA) (502), acid end-capped Resomer RG 502H PLGA (502H), and a 502H:502 mixture (3:1) via hot melt extrusion (HME). The prepared intravitreal implants (20 and 40% DM loaded in each PLGA) were thoroughly investigated to determine the effect of different end-capped PLGA and drug loading on the long-term release profile of DM. The implants were characterized for solid-state active pharmaceutical ingredient (APIs) using DSC and SWAXS, water uptake during stability study, the crystal size of API in the implant matrix using hot-stage polarized light microscopy, and in vitro release profile. The kinetics of PLGA release was thoroughly investigated using quantitative 1H NMR spectroscopy. The polymorph of DM crystal was found to remain unchanged after the extrusion and stability study. However, around 3 times reduction in API particle size was observed after the HME process. The morphology and content uniformity of the RT-stored samples were found to be comparable to the initial implant samples. Interestingly, the samples (mainly 502H) stored at 40 °C and 75% RH for 30 d demonstrated marked deformation and a change in content uniformity. The rate of DM release was higher in the case of 502H samples with a higher drug loading (40% w/w). Furthermore, a simple digital in vitro DM release profile derived for the formulation containing a 3:1 ratio of 502H and 502 was comparable with the experimental release profile of the respective polymer mixture formulation. The temporal development of pores and/or voids in the course of drug dissolution, evaluated using µCT, was found to be a precursor for the PLGA release. Overall, the release profile of DM was found to be dependent on the PLGA type (independent of subtle changes in the formulation mass and diameter). However, the extent of release was found to be dependent on DM loading. Thus, the present investigation led to a thorough understanding of the physicochemical properties of different end-capped PLGAs and the underlying formulation microstructure on the release profile of a crystalline water-insoluble drug, DM, from the PLGA-based implant.

Mechanoactivation as a Tool to Assess the Autoxidation Propensity of Amorphous Drugs.

Mechanoactivation has attracted considerable attention in the pharmaceutical sciences due to its ability to generate amorphous materials and solid-state synthetic products without the use of solvent. Although some studies have reported drug degradation during milling, no studies have systematically investigated the use of mechanoactivation in predicting drug degradation in the solid state. Thus, this work explores the autoxidation of drugs in the solid state by comilling amorphous mifepristone (MFP):polyvinylpyrrolidone vinyl acetate (PVPVA) and amorphous olanzapine (OLA):PVPVA. MFP was amorphized by ball milling and OLA by quench cooling techniques. Subsequently, comilling the amorphous drugs in the presence of a 10-fold weight ratio of PVPVA (the excipient containing reactive free radicals) was performed at several milling frequencies to identify the kinetics of mechano-autoxidation over milling durations. Overall, milling led to the degradation of up to 5% drug in the solid state. The autoxidation mechanism was confirmed by performing a stress study in the solution at 50 °C for 5 h, by using a 10 mM azo-bis(isobutyronitrile) (AIBN) as a stressing agent. By deconvoluting the effect of milling frequency and the energy on the extent and kinetics of milling-induced autoxidation of amorphous drugs, it was possible to fit an extended Arrhenius model that allowed extrapolation of mechanoactivated degradation rates (Km) to zero milling frequencies. Further, the autoxidation rates of drugs stored at high temperatures were observed to follow an Arrhenius behavior. A good degree of agreement was observed between the model predictions obtained by mechanoactivation (Km) to the reaction rates observed under accelerated temperatures. Additionally, the impact of adding an antioxidant (e.g., butylated hydroxytoluene) to the mixture during comilling was also examined. This study can be helpful in evaluating the stability of amorphous solids stored in accelerated (non-hermetic) conditions, in screening solid-state autoxidation propensity of drugs, and for the rational selection of antioxidants.

Temperature cycling-induced formation of crystalline coatings.

The surface of particles is the hotspot of interaction with their environment and is therefore a major target for particle engineering. Particles with tailored coatings are greatly desired for a range of different applications. Amorphous coatings applied via film coating or microencapsulation have frequently been described in the pharmaceutical context and usually result in homogeneous surfaces. In the present study we have been exploring the feasibility of coating core particles with crystalline substances, a matter that has rarely been investigated. The expansion of the range of possible coating materials to include small organic molecules enables completely new product properties to be achieved. We present an approach based on temperature cycles performed in a tubular crystallizer to result in engineered crystalline coatings on excipient core particles. By manipulating the process settings and by the choice of coating substance we are able to tailor surface roughness, topography as well as surface chemistry. Benefits of our approach are demonstrated by using resulting particles as carriers in dry-powder-inhaler formulations. Depending on the resulting surface chemistry and surface roughness, coated carrier particles show varying fitness for delivering the model API salbutamol sulphate to the lung.

Machine Learning-Enabled NIR Spectroscopy. Part 2: Workflow for Selecting a Subset of Samples from Publicly Accessible Data.

An increasingly large dataset of pharmaceutics disciplines is frequently challenging to comprehend. Since machine learning needs high-quality data sets, the open-source dataset can be a place to start. This work presents a systematic method to choose representative subsamples from the existing research, along with an extensive set of quality measures and a visualization strategy. The preceding article (Muthudoss et al.. in AAPS PharmSciTech 23, 2022) describes a workflow for leveraging near infrared (NIR) spectroscopy to obtain reliable and robust data on pharmaceutical samples. This study describes the systematic and structured procedure for selecting subsamples from the historical data. We offer a wide range of in-depth quality measures, diagnostic tools, and visualization techniques. A real-world, well-researched NIR dataset was employed to demonstrate this approach. This open-source tablet dataset ( http://www.models.life.ku.dk/Tablets ) consists of different doses in milligrams, different shapes, and sizes of dosage forms, slots in tablets, three different manufacturing scales (lab, pilot, production), coating differences (coated vs uncoated), etc. This sample is appropriate; that is, the model was developed on one scale (in this research, the lab scale), and it can be great to investigate how well the top models are transferable when tested on new data like pilot-scale or production (full) scale. A literature review indicated that the PLS regression models outperform artificial neural network-multilayer perceptron (ANN-MLP). This work demonstrates the selection of appropriate hyperparameters and their impact on ANN-MLP model performance. The hyperparameter tuning approaches and performance with available references are discussed for the data under investigation. Model extension from lab-scale to pilot-scale/production scale is demonstrated. HIGHLIGHTS: • We present a comprehensive quality metrics and visualization strategy in selecting subsamples from the existing studies • A comprehensive assessment and workflow are demonstrated using historical real-world near-infrared (NIR) data sets • Selection of appropriate hyperparameters and their impact on artificial neural network-multilayer perceptron (ANN-MLP) model performance • The choice of hyperparameter tuning approaches and performance with available references are discussed for the data under investigation • Model extension from lab-scale to pilot-scale successfully demonstrated.

Trends in oral small-molecule drug discovery and product development based on product launches before and after the Rule of Five.

In 1997, the 'Rule of Five' (Ro5) suggested physicochemical limitations for orally administered drugs, based on the analysis of chemical libraries from the early 1990s. In this review, we report on the trends in oral drug product development by analyzing products launched between 1994 and 1997 and between 2013 and 2019. Our analysis confirmed that most new oral drugs are within the Ro5 descriptors; however, the number of new drug products of drugs with molecular weight (MW) and calculated partition coefficient (clogP) beyond the Ro5 has slightly increased. Analysis revealed that there is no single scientific or technological reason for this trend, but that it likely results from incremental advances are being made in molecular biology, target diversity, drug design, medicinal chemistry, predictive modeling, drug metabolism and pharmacokinetics, and drug delivery.