Comparative evaluation of the powder properties and compression behaviour of a new cellulose-based direct compression excipient and Avicel PH-102.

This study compares the compression behaviour of a new cellulose-based tableting excipient, hereinafter referred to as UICEL-A/102, and Avicel PH-102, a commercial direct compression excipient commonly referred to as microcrystalline cellulose (MCC). UICEL-A/102 shows the cellulose II lattice, while Avicel PH-102 belongs to the cellulose I polymorphic form. The median particle diameters of UICEL-A/102 and Avicel PH-102 fractions used in the study were 107 and 97 microm, respectively. Compared with Avicel PH-102, UICEL-A/102 was more dense; the relative poured and tapped densities were: 0.277 and 0.327 (vs 0.195 and 0.248 for Avicel PH-102), respectively. The true density, rhotrue, of the two materials was comparable ( approximately 1.56 g cm(-3)). The slopes of the in-die and out-of-die Heckel curves for Avicel PH-102 were steeper than for UICEL-A/102. The relative density versus applied pressure plot was in good agreement with the modified Heckel equation. The out-of-die and in-die minimal pressure susceptibility (chipmin) values calculated were 3.36 x 10(-3) and 8.09 x 10(-3) MPa(-1) for UICEL-A/102 and 8.00 x 10(-3) and 16.12 x 10(-3) MPa(-1) for Avicel PH-102, respectively. The elastic recovery profiles showed UICEL-A/102 to be more elastic than Avicel PH-102. In conclusion, UICEL-A/102 and Avicel PH-102 differ in their compression behaviour under pressure. The different polymorphic forms could provide a possible explanation.

Concentrate feeding strategy in lactating dairy cows: metabolic and endocrine changes with emphasis on leptin.

This study aimed to evaluate metabolic and endocrine adaptations to energy intake in multiparous Holstein cows (n = 90; mean 9434 kg energy-corrected milk yield/305 d) over the first 20 wk postpartum and to assess the association of leptin with metabolic, endocrine, and zootechnical traits. Concentrates were fed automatically for 24 h at 30% (C30) or 50% (C50) of total dry matter intake (DMI) from wk 1 to 10 postpartum and at linearly reduced amounts thereafter. Roughage was fed for ad libitum intake. The DMI was measured over 24 h; milk yield and body weight (BW), twice/d; milk composition, 4 times/wk; and milk acetone, weekly. Blood samples for determination of metabolite, hormone, and electrolyte concentrations and enzyme activities were obtained at wk 2 prepartum, and at wk 1 to 16 and at wk 20 postpartum from 0730 to 0900. Body condition scores (BCS) and backfat thickness were measured postpartum and during wk 1, 4, 8, 12, 16, and 20. Energy balance (EB) was considerably lower, but milk yield only slightly lower, in C30 than C50. Metabolic stress was more marked in C30 than C50, expressed by lower, glucose, insulin, insulin-like growth factor-1 (IGF-1), triiodothyronine, milk protein, and lactose concentrations, higher nonesterified fatty acid, beta-hydroxybutyrate, growth hormone, and milk acetone concentrations, and an accelerated decrease in BCS and backfat thickness. Nevertheless, C30 adapted successfully and thus maintained high milk yields despite negative EB. Leptin concentrations were lower in C30 than in C50 over the first 20 wk postpartum and were positively associated with BCS, EB, BW, cholesterol, albumin, insulin, and IGF-1; negatively associated with DMI and triiodothyronine; and were higher in cows calving in spring than in fall. Leptin is one among several factors involved in the regulation of energy metabolism and may be important for overall homeostatic and homeorhetic control of metabolism and thus for maintenance of performance.

Estimation of energy balance at the individual and herd level using blood and milk traits in high-yielding dairy cows.

This study aimed to estimate individual and herd-level energy balance (EB) using blood and milk traits in 90 multiparous high-yielding Holstein cows, held on a research farm, from wk 1 to 10 postpartum (p.p.) and to investigate the precision of prediction with successively decreased data sets simulating smaller herd sizes and with pooled samples. Dry matter intake, milk yield, and BW were measured daily from parturition through wk 10 p.p. Milk composition was determined 4 times per week, and milk acetone was measured weekly. Blood samples for the determination of metabolites, hormones, electrolytes, and enzyme activities were taken weekly from wk 1 to 10 p.p. between 0730 and 0900. Body condition scores and ultrasonic measurements of backfat thickness and fat depth in the pelvic area were evaluated in wk 1, 4, and 8 p.p. Concentrations of glucose, cholesterol, urea, insulin, insulin-like growth factor-1, triiodothyronine, and thyroxine (T4) in blood plasma and of lactose and urea in milk were positively correlated with EB, whereas concentrations of nonesterified fatty acids (NEFA), creatinine, albumin, beta-hydroxybutyrate, and growth hormone and enzyme activities in blood, and concentrations of fat, protein, fat:lactose ratio, and acetone in milk were negatively correlated with EB. Leptin concentration was not correlated to EB over the first 10 wk p.p. To estimate EB linear mixed-effects, models were developed by backward selection procedures. The most informative traits for estimation of EB were the fat:lactose ratio in milk and NEFA and T4 concentrations in blood. The precision of estimation of EB in individual cows was low. Using blood in addition to milk traits did not result in higher precision of estimation of herd-level EB, and decreasing sample sizes considerably lowered the precision of EB prediction. Estimation of overall mean herd-level EB over the first 10 wk p.p. using pooled samples was precise even with small sample sizes, but does not consider the level of EB in particular weeks. In conclusion, estimation of herd-level EB at individual weeks using milk traits only has practical implication with herd sizes of > or = 100 cows if calving is highly seasonal and of or = 400 cows if calving is uniformly distributed. Using blood in addition to milk traits does not improve precision of estimation of herd-level EB, regardless of sample size.

New trends in the production of pharmaceutical granules: the classical batch concept and the problem of scale-up.

The moist agglomeration process, i.e. the wet massing, screening, and subsequent drying is often a critical unit operation. The correct amount of granulating liquid and the correct monitoring and detection of the granulation kinetics are important issues. The method to monitor the kinetics needs to be robust and should be applicable for any batch size. In this context, the theory of scale-up and the monitoring of the moist agglomeration process are reviewed. It has to be kept in mind that the production of granules in the pharmaceutical industry is still based on a batch concept. This concept offers many advantages with respect to quality assurance as a batch can be accepted or rejected. From experience, it is well known, however, that the scale-up of the batch size may lead to problems. This fact is due to the variety of the equipment involved and to the fact that there is a lack of well-known 'scale-up invariant' parameters. A survey of the granulation end-point detection procedure shows that the majority of the equipment manufacturers offer mixer/kneaders for the moist agglomeration process instrumented with a power consumption device. In this review, this and other approaches are discussed and emphasis is placed on how to best use the power consumption method. The question of robust formulations leads to the conclusion that, for a robust dosage form design, new concepts such as percolation theory have to be applied. A typical example is presented, which illustrates the effect of a percolation phenomenon.

New trends in the production of pharmaceutical granules: batch versus continuous processing.

In the pharmaceutical industry, the production of granules is based on a batch concept. This concept offers many advantages with respect to quality assurance as a batch can be accepted or rejected. However, the scale-up of the batch size may lead to problems. The variety of the equipment involved often does not facilitate the scale-up process. In order to avoid scale-up problems, continuous or semi-continuous processes have to be evaluated as alternatives to a batch production. Thus, a quasi-continuous production line is presented, which permits the production of small-scale batches, e.g. for clinical trials and for large-scale batches using the same equipment.

Evaluation of flow properties of dry powder inhalation of salbutamol sulfate with lactose carrier.

The effects of the flow and packing properties of a drug/carrier powder mixture on emission of drug adhering to the carrier from capsules and inhalation devices were investigated. Model powder mixtures were designed consisting of lactose carriers with different particle shapes were prepared by surface treatment and micronized salbutamol sulfate. These powder mixtures were aerosolized by a Spinhaler, and in vitro deposition properties of salbutamol sulfate were evaluated by a twin impinger. The flow properties of the mixed powders were evaluated by the Carr's flowability index (FI) and Hausner's ratio (HR). The packing properties of the mixed powders were determined employing the tapping method. Compared with the powder mixed with the untreated lactose carrier, the FI, HR, and the constant K in Kawakita's equation of the powder mixture prepared using the surface-treated lactose carrier were significantly different, showing that the flow and packing properties of the drug/carrier powder mixture were improved. Using this surface-treated system, the handling of the powder mixture when packing into capsules is improved, which is desirable for handling dry powder inhalants. The fraction (%) of drug emitted from capsules and devices (EM) and the FI of the powder mixture were correlated. As the flow properties improved, the outflow of the powder mixture from capsules and devices became easier, and emission of drug adhering on the carrier from capsules and devices improved. Improvement of the inhalation process, such as the drug particles emitted from the inhalation system, is valuable for increasing inhalation properties of dry powder inhalation.

A novel approach to the characterization of polar liquids. Part 1: pure liquids.

Liquid dosage forms, generally based on aqueous solutions, take an important role in drug administration. The approaches to a theoretical description of solvent and solubility properties have not yet proved completely satisfying. In this work, the Debye equation, which describes well polar and nonpolar molecules in an ideal gas, is extended to liquids. For this purpose, the Debye equation was modified and the term (E(i)/E) was introduced (E(i)=internal electric field, E=applied external electric field). Pure polar and nonpolar solvents were measured between 290.7 and 343.2 K. The values of (E(i)/E) were compared with the correlation factor g of the Kirkwood-Fröhlich equation, a measure for molecular pair correlations. For polar solvents, the relationship E(i)/E=m(1/T)+b as a function of temperature T was found. Associating compounds showed negative values of (E(i)/E) with a strong temperature dependency; the latter can be expressed by the slope m. A correlation between the absolute value of m and the corresponding Hildebrand solubility parameter delta could be established. This new approach allows to describe polar hydrogen-bonding liquids and provides a tool for a more rational design of liquid dosage forms.

Biosynthesis of lipstatin. Incorporation of multiply deuterium-labeled (5Z,8Z)-tetradeca-5,8-dienoic acid and octanoic acid.

Fermentation experiments with Streptomyces toxytricini were performed using (5Z,8Z)-[10,11,12,12-(2)H]tetradeca-5,8-dienoic acid or a mixture of [2,2-(2)H(2)]- and [8,8,8-(2)H(3)]octanoic acid as supplements. (2)H NMR and mass spectroscopy confirmed the incorporation of (5Z,8Z)-[10,11,12,12-(2)H]tetradeca-5,8-dienoic acid into the C(13) side chain as well as into the C(6) side chain of lipstatin. Moreover, deuterium was incorporated into the C(6) side chain of lipstatin from the 8-position but not from the 2-position of octanoate. The data establish that the beta-lactone moiety of lipstatin is formed by condensation of a C(8) and a C(14) fatty acid with a concomitant exchange of the H-2 atoms of the C(8) fatty acid.

Thermal sterilization of heat-sensitive products using high-temperature short-time sterilization.

High-temperature short-time (HTST) sterilization with a continuous-flow sterilizer, developed for this study, was evaluated. The evaluation was performed with respect to (a) the chemical degradation of two heat-sensitive drugs in HTST range (140-160 degrees C) and (b) the microbiological effect of HTST sterilization. Degradation kinetics of two heat-sensitive drugs showed that a high peak temperature sterilization process resulted in less chemical degradation for the same microbiological effect than a low peak temperature process. Both drugs investigated could be sterilized with acceptable degradation at HTST conditions. For the evaluation of the microbiological effect, Bacillus stearothermophilus ATCC 7953 spores were used as indicator bacteria. Indicator spore kinetics (D(T), z value, k, and E(a)), were determined in the HTST range. A comparison between the Bigelow model (z value concept) and the Arrhenius model, used to describe the temperature coefficient of the microbial inactivation, demonstrated that the Bigelow model is more accurate in prediction of D(T) values in the HTST range. The temperature coefficient decreased with increasing temperature. The influence of Ca(2+) ions and pH value on the heat resistance of the indicator spores, which is known under typical sterilization conditions, did not change under HTST conditions.

The use of fluorescence resonance energy transfer to study the disintegration kinetics of liposomes containing lysolecithin and oleic acid in rat plasma.

PURPOSE: To validate Fluorescence Resonance Energy Transfer (RET) as method to monitor disintegration of fluorescently labeled liposomes varying in lysolecithin/oleic acid (equimolar) content, lysolecithin fatty acid composition and vesicle size in rat blood plasma and buffer. METHODS: NBD-PE and Rho-PE were used for RET. The measurements were performed on a Perkin Elmer LS-50 spectrofluorimeter. Liposomes were prepared by the extrusion method. RESULTS: Analysis of the RET data was optimised using a fitting procedure to correct for fluorescence interference by plasma. The disintegration patterns of liposomes could be described by a biexponential decay model. Disintegration rate increased at increasing lysolecithin/oleic acid content and decreasing size. In contrast, all liposomes showed no disintegration in buffer. CONCLUSIONS: RET is a suitable method to monitor liposome disintegration in non-diluted plasma. Rate and extent of liposome disintegration increases at decreasing liposome size and increasing lysolecithin/oleic acid content.

Meal patterns and meal-induced metabolic changes in calves fed milk ad lib.

The feeding behavior of 11 calves fed milk ad lib was characterized and analyzed at the age of 5 weeks, and the short-term changes in the plasma concentrations of various metabolites (glucose, lactate, free fatty acids, triglycerides, beta-hydroxybutyrate) and insulin in relation to a representative spontaneous milk meal were measured during the following week. In a 6-day period, the calves consumed 287 (=86%) of a total of 335 milk meals during the light phase from 0500-2200 [on average, 4.4 +/- 0.5 (mean +/- SEM) meals]. The meal size and duration during light were 2.0 +/- 0.3 kg and 5.3 +/- 0.3 min, respectively. However, only 0.7 +/- 0.1 milk meals of similar size and duration were consumed during the dark phase. The plasma concentrations of insulin and glucose increased in response to the spontaneous milk meal and remained elevated for at least 2 h after meal end. The plasma concentrations of triglycerides, free fatty acids, and beta-hydroxybutyrate also increased after meal termination, and remained elevated until 40 min (triglycerides, free fatty acids) and 60 min (beta-hydroxybutyrate) after meal end, respectively. The observed spontaneous milk intake patterns were similar to the natural suckling behavior described for calves, suggesting that the conditions of the present experiment did not disrupt the animals' natural feeding behavior. Some of the profound metabolic changes in relation to a spontaneous milk meal might contribute to the control of milk intake in calves, but further experiments are necessary to test this idea.

A new theoretical approach to tablet strength of a binary mixture consisting of a well and a poorly compactable substance.

The objective of this study was to analyse the tensile strength of a well and a poorly compactable substance in a tablet mixture. Recent developments in the theory of percolation were taken into account and two power laws are proposed, one for the tensile strength as a function of the relative density of the mixture, and the other for the relationship between the strength and compaction pressure. Both equations are assumed to be valid in a comparatively low pressure range. A universal testing instrument Zwick UPM 1478 was used for the manufacture and testing of the compacts. Mixtures of Avicel PH101 and paracetamol at different ratios were chosen as model systems. The experimental results showed that the proposed model equations fitted the experimental data reasonably well for all mixture ratios. It was observed that the critical solid fraction of the mixture, i.e. the strength percolation threshold, increased with rising amounts of the drug. We investigated the strength threshold not only in terms of the solid fraction, but also in terms of the mass fraction (excipient percolation threshold). It is assumed that a tablet can only be produced with a certain minimal amount of the well compactable substance that is needed to build a percolating cluster in the tablet. An interpretation is therefore provided for the dilution capacity of a direct tableting excipient with a poorly compactable drug. The dilution capacity was experimentally determined according to the method of Minchom and Armstrong (Br. Pharm. Conf. (1987) 69 pp.). Our experimental estimate of 79.9% drug is in perfect agreement with our proposed theoretical calculation of 79.7%. These estimates are, however, much higher than the one reported in a recent study (Y. Habib, L. Augsburger, G. Reier, Th. Wheatley, R. Shangraw, Dilution potential: a new perspective, Pharm. Dev. Tech. 1 (2) (1996) 205-212) where the dilution capacity of the same mixture was investigated. This discrepancy can be explained based on the different pressure ranges and extrapolation techniques that were used. As a conclusion, concepts of the percolation theory can successfully be applied to the kind of mixture studied in this paper. It is conceivable that the theoretical tools presented can also be applied to mixtures of more than two substances if they consist of a single well compactable excipient and several poorly compactable components. Such mixtures are relevant for the development of direct compressible tableting formulations.

[Effect of mixing of fine carrier particles on dry powder inhalation property of salbutamol sulfate (SS)].

The most commonly used formulations for dry powder inhalations are binary ordered mixes composed of micronized drugs and coarse carriers. An optimal dry powder aerosol formulation should possess an optimal inhalation property and a good flow property. These characteristics are especially important for a multidose dry powder inheler (DPI). In the present study, model powder blend were prepared consisting of synthesized sugar (different particle sized isomalt; IM-PF, IM-FS, IM-F) as a carrier and micronized salbutamol sulfate (SS). These ordered mixtures were aerosolized by the multidose JAGO DPI (SkyePharma AG) and in vitro deposition properties (fine particle fraction, FPF) were evaluated by a twin impinger (TI) at a flow rate of 60 l/min. The separation property between SS and carrier particles was investigated by the centrifuge method and air jet sieve (AJS) method. It was found that FPF decreased with increasing carrier particle size. However, a large carrier particle possesses a good flow property. Therefore, the effect of mixing of fine carrier particles (IM-PF) into the large carrier particles (IM-FS) on dry powder inhalation property was investigated. When the proportion of IM-PF (fine carrier) increase from 0% to 25% of the total carrier powder blend, the FPF also increases from 16.7% to 38.9%. It is concluded that the effect of mixing of fine carrier particles might be a suitable method for improving the dry powder inhalation properties.

Flow-cytometric investigation of cellular metabolism during oxidative stress and the effect of tocopherol.

Many studies and scientific publications report on potentially beneficial effects of the lipophilic anti-oxidant vitamin E on cellular metabolic pathways. The present work presents data on the influence of tocopherol on different intracellular parameters of intact and living human skin fibroblasts by flow-cytometric measurements. The parameters analysed were the intracellular pH, representing cell metabolism and cell function, intracellular glutathione, representing one of the cell's own radical scavenger enzyme systems, membrane potential and cell viability. In order to cause large numbers of free radicals cells were UVB-irradiated prior to measurement. The results of the flow-cytometric measurements indicate that vitamin E has significant protecting effects on the measured biochemical parameters during oxidative stress. In the presence of the lipophilic radical scavenger a significant stabilizing effect on pH, intracellular glutathione levels and membrane potential could be observed. Furthermore, vitamin E administration was associated with increased cell viability after UVB irradiation.

[Effect of separation characteristics between salbutamol sulfate (SS) particles and model carrier excipients on dry powder for inhalation].

Most often dry powder for inhalation are formulated as ordered mixtures of a carrier excipient and a micronized drug substance. In the present study, model powder blends were prepared from a mixture of lactose alpha-monohydrate, micro-crystalline cellulose pellets or synthesized sugar as carrier particles, and micronized salbutamol sulfate (SS). These ordered mixtures were aerosolized by the multidose JAGO dry powder inhaler (DPI) and their in vitro deposition properties were evaluated by a twin impinger (TI). The separation force between SS particles and carrier particles was investigated by the centrifuge method. In addition, the use of the air jet sieve (AJS) method was investigated to assess the separation behavior of drug particles from carrier excipient. Powder blends were sieved through a 325 mesh wire screen of an air jet sieve at an air pressure of 1500 Pa. The amount of drug deposited at the carrier surface was analysed before and after the sieving to calculate the percentage of the drug retained. A relationship was found between in vitro deposition properties (fine particle fraction, FPF) and the separation characteristics obtained by the centrifuge method and by the AJS method. The AJS method might be a suitable alternative for evaluating separation of a drug particle from carrier particles and hence can be used for the formulation screening of the dry powder inhalation.

Percolation theory, conductivity and dissolution of hydrophilic suppository bases (PEG systems).

We investigated the conductivity of binary mixtures of different volume to volume ratios of liquid polyethylene glycol 200 (PEG 200) and solid PEG 6000. The conductivity was measured using a specially designed cell, which could be filled with the sample to be analysed. The results were analysed by the help of the percolation theory and its basic equation: X=S(p-p(c))(q) with X=gamma the conductivity and X=k the dissolution rate constant of binary mixtures with p the concentration of PEG 200 including Mepyraminmaleat as a marker substance. The intrinsic dissolution rate was determined spectrophotometrically. The simultaneous determination of the critical exponent q and the percolation threshold p(c) of the conductivity experiment yielded the following values q=1.83+/-0.05 and p(c)=11.6+/-1.1% with a squared correlation coefficient R(2)=0. 9986. For q=mu=2.00 fixed the percolation threshold is equal to p(c)=8.4+/-1.3% with R(2)=0.9984. Thus if the same percolation threshold p(c)=8.4+/-1.3% is adopted for the dependence of the dissolution process q becomes equal to q approximately mu=1. 94+/-0.045. R(2)=0.9991. This result is in excellent agreement with the theoretical prediction that the permeability of a porous network scales in the same way as the conductivity.

Fracture in disordered media and tensile strength of microcrystalline cellulose tablets at low relative densities.

The purpose of this study is to establish a theoretical basis for the tensile strength of low density tablets. In a first step, a lattice model based on percolation theory is presented. As a theoretical result, a power law is obtained for the lattice strength. The exponent in this law is expected to be universal and as a numerical value Tf congruent with 2.7 is proposed. The result is identical with an earlier theoretical finding from an alternative approach proposed by. In a second step, the new model equation is applied to the tensile strength of low density tablets. The compacts were manufactured and tested with an universal testing instrument Zwick UPM 1478 (Zwick-Roell). Different types of microcrystalline cellulose Emcocel 50M, Emcocel 90M, Avicel PH101 and Avicel PH102 were assayed as model excipients because of their ability to form tablets at comparatively low relative densities (rhor). For determination of the tensile strength, two different strain rates 0.5 and 25 mm min-1 were examined. All experimentally determined exponents were in the same range with an average of Tf=3.2+/-0.1 and the critical solid fractions (rhorc), yielded values close the relative bulk densities. In a third step, the new model is compared to the Ryshkewitch-Duckworth equation. This exponential relationship of the tensile strength and porosity was found to have an inferior fitting adequacy than the new power law. As a conclusion, the lattice model presented is able to explain the power law behaviour of the tensile strength as a function of the relative density with an exponent close to three. The expected universal character of this exponent was supported by the results of the assayed substances at two different strain rates. Plus, in the case of the tested substances, the new relationship between the tensile strength and the relative density should be preferred to the often used exponential function. However, further studies have to be conducted to know more about the validity of the new model.

Pressure susceptibility of polymer tablets as a critical property: a modified Heckel equation.

The pressure susceptibility (chip), which is defined as the decrease of porosity (epsilon) under pressure was investigated. Of special interest are compacts obtained at very low pressures, because of the transition between the state of a powder and the state of a tablet. This range was found to be critical in respect to a diverging pressure susceptibility. Above a critical porosity (epsilonc) or below the corresponding relative density (rhoc), no pressure susceptibility can be defined, because of no rigid structure exists. To take this into account, a simple function was proposed for the pressure susceptibility: chip approximately 1/(epsilonc - epsilon). This proposal leads to a new porosity vs pressure relationship. The new model was compared to the Heckel equation that involves a constant pressure susceptibility. Various polymers were tested from "out of die" measurements, and the new relationship was found superior to the Heckel equation. As a conclusion, the pressure susceptibility exhibits a curvature that can be called critical at low relative densities. Consequently, a better understanding evolves as to why the Heckel equation is not valid at low pressures. The new model has proven to be adequate for polymer tablets but, so far it is not clear whether other substances exhibit the same performance. Especially tableting materials exhibiting brittle fracture will be of interest considering their importance in compaction technology.

Advantages of Artificial Neural Networks (ANNs) as alternative modelling technique for data sets showing non-linear relationships using data from a galenical study on a solid dosage form.

Artificial Neural Networks (ANN) methodology was used to assess experimental data from a tablet compression study showing highly non-linear relationships (i.e. measurements of ejection forces) and compared to classical modelling technique (i.e. Response Surface Methodology, RSM). These kinds of relationships are known to be difficult to model using classical methods. The aim of this investigation was to quantitatively describe the achieved degree of data fitting and predicting abilities of the developed models. The comparison between the ANN and RSM was carried out both graphically and numerically. For comparing the goodness of fit, all data were used, whereas for the goodness of prediction the data were split into a learning and a validation data set. Better results were achieved for the model using ANN methodology with regard to data fitting and predicting ability. All determined ejection properties were mainly influenced by the concentration of magnesium stearate and silica aerogel, whereas the other factors showed very much lower effects. Important relationships could be recognised from the ANN model only, whereas the RSM model ignored them. The ANN methodology represents a useful alternative to classical modelling techniques when applied to variable data sets presenting non-linear relationships.

Pitfalls of artificial neural networks (ANN) modelling technique for data sets containing outlier measurements using a study on mixture properties of a direct compressed dosage form.

An application of the Artificial Neural Networks (ANN) methodology was investigated using experimental data from a mixture properties study and compared to classical modelling technique (i.e. Response Surface Methodology, RSM) both graphically and numerically. The aim of this investigation was to quantitatively describe the achieved degree of data fitting and robustness of the developed models. For comparing the goodness of fit, the R2 coefficient was used, whereas for the robustness of the models an outlier measurement was integrated in the data set. Comparable results were achieved for both ANN- and RSM methodologies for data fitting. The robustness of the models towards outliers was clearly better for the RSM methodology. All determined mixture properties were mainly influenced by the concentration of silica aerogel, whereas the other factors showed very much lower effects. For that reason the physical properties of this excipient (e.g. its specific surface area) are of importance for the behaviour of the mixtures.