Modelling formulations using gene expression programming--a comparative analysis with artificial neural networks.

This study has investigated the utility and potential advantages of gene expression programming (GEP)--a new development in evolutionary computing for modelling data and automatically generating equations that describe the cause-and-effect relationships in a system--to four types of pharmaceutical formulation and compared the models with those generated by neural networks, a technique now widely used in the formulation development. Both methods were capable of discovering subtle and non-linear relationships within the data, with no requirement from the user to specify the functional forms that should be used. Although the neural networks rapidly developed models with higher values for the ANOVA R(2) these were black box and provided little insight into the key relationships. However, GEP, although significantly slower at developing models, generated relatively simple equations describing the relationships that could be interpreted directly. The results indicate that GEP can be considered an effective and efficient modelling technique for formulation data.

Advantages of neurofuzzy logic against conventional experimental design and statistical analysis in studying and developing direct compression formulations.

This study has investigated the utility and potential advantages of an artificial intelligence technology - neurofuzzy logic - as a modeling tool to study direct compression formulations. The modeling performance was compare with traditional statistical analysis. From results it can be stated that the normalized error obtained from neurofuzzy logic was lower. Compared to the multiple regression analysis neurofuzzy logic showed higher accuracy in prediction for the five outputs studied. Rule sets generated by neurofuzzy logic are completely in agreement with the findings based on statistical analysis and advantageously generate understandable and reusable knowledge. Neurofuzzy logic is easy and rapid to apply and outcomes provided knowledge not revealed via statistical analysis.

Data mining of fractured experimental data using neurofuzzy logic-discovering and integrating knowledge hidden in multiple formulation databases for a fluid-bed granulation process.

In the pharmaceutical field, current practice in gaining process understanding by data analysis or knowledge discovery has generally focused on dealing with single experimental databases. This limits the level of knowledge extracted in the situation where data from a number of sources, so called fractured data, contain interrelated information. This situation is particularly relevant for complex processes involving a number of operating variables, such as a fluid-bed granulation. This study investigated three data mining strategies to discover and integrate knowledge "hidden" in a number of small experimental databases for a fluid-bed granulation process using neurofuzzy logic technology. Results showed that more comprehensive domain knowledge was discovered from multiple databases via an appropriate data mining strategy. This study also demonstrated that the textual information excluded in individual databases was a critical parameter and often acted as the precondition for integrating knowledge extracted from different databases. Consequently generic knowledge of the domain was discovered, leading to an improved understanding of the granulation process.

Detection, Quantification, and Vegetative Compatibility of Verticillium dahliae in Potato and Mint Production Soils in the Columbia Basin of Oregon and Washington.

Soil samples from 87 fields intended for potato production in the Columbia Basin of Washington and Oregon and 51 fields intended for mint production in Washington were assayed on a semiselective medium to quantify populations of Verticillium dahliae. The pathogen was isolated from 77 (89%) of the fields intended for potato production and 41 (80%) of the fields intended for mint production. Population densities ranged from 0 to 169 propagules/g of air-dried soil in fields intended for potato production and 0 to 75 propagules/g of air-dried soil in fields intended for mint production. Isolates of V. dahliae were recovered from soil assay plates and pure cultures were prepared to provide a collection of isolates for vegetative compatibility analysis. Among fields intended for potato production from which isolates of the fungus were assigned to a vegetative compatibility group (VCG), 93% of the fields were found to contain one or more isolates assigned to VCG 4A, nine (23%) contained one or more isolates assigned to VCG4B, and only one (3%) contained any isolates assigned to VCG 2B. In the case of fields planned for mint production in 1999 from which isolates of the fungus were assigned to a VCG, 13 fields (81%) were found to contain one or more isolates assigned to VCG 4A, 7 (44%) contained one or more isolates assigned to VCG 4B, and 5 (31%) contained one or more isolates assigned to VCG 2B. VCG 4A isolates of V. dahliae are widespread and numerous, particularly following potato production, but cause only mild to moderate symptoms in mint; therefore, this pathotype is unlikely to seriously endanger subsequent plantings of mint. However, planting potato in a field recently used to produce mint may pose a significant risk to the potato crop if high populations of the VCG4A pathotype (highly aggressive to potato) predominate. Preplant assessment of soil populations of V. dahliae without regard for the relative populations of various pathotypes present in a particular sample may lead to information not fully useful in integrated pest management systems.

Comparison of neurofuzzy logic and decision trees in discovering knowledge from experimental data of an immediate release tablet formulation.

Understanding of the cause-effect relationships between formulation ingredients, process conditions and product properties is essential for developing a quality product. However, the formulation knowledge is often hidden in experimental data and not easily interpretable. This study compares neurofuzzy logic and decision tree approaches in discovering hidden knowledge from an immediate release tablet formulation database relating formulation ingredients (silica aerogel, magnesium stearate, microcrystalline cellulose and sodium carboxymethylcellulose) and process variables (dwell time and compression force) to tablet properties (tensile strength, disintegration time, friability, capping and drug dissolution at various time intervals). Both approaches successfully generated useful knowledge in the form of either "if then" rules or decision trees. Although different strategies are employed by the two approaches in generating rules/trees, similar knowledge was discovered in most cases. However, as decision trees are not able to deal with continuous dependent variables, data discretisation procedures are generally required.

Investigation of an artificial intelligence technology--Model trees. Novel applications for an immediate release tablet formulation database.

This study has investigated an artificial intelligence technology - model trees - as a modelling tool applied to an immediate release tablet formulation database. The modelling performance was compared with artificial neural networks that have been well established and widely applied in the pharmaceutical product formulation fields. The predictability of generated models was validated on unseen data and judged by correlation coefficient R(2). Output from the model tree analyses produced multivariate linear equations which predicted tablet tensile strength, disintegration time, and drug dissolution profiles of similar quality to neural network models. However, additional and valuable knowledge hidden in the formulation database was extracted from these equations. It is concluded that, as a transparent technology, model trees are useful tools to formulators.

The application of molecular modelling to the interpretation of inverse gas chromatography data.

The use of molecular modelling in the interpretation of inverse gas chromatography data is discussed. Crystal faces can be visualised and likely cleavage planes calculated using the surface attachment energies. Assuming that the preferred cleavage plane is the crystal face with the smallest attachment energy then the predominant crystal faces of a crystalline particle can be predicted. Surface adsorption can be modelled using Van der Waals and electrostatic interactions to evaluate the interaction energies between individual atoms of the probe molecule and atoms of the test molecule orientated as in the surface. Using examples of pharmaceutical materials, modelling has been shown to be successful in the understanding of changes in the surface energetics.

The evaluation of the rheological properties of lactose/microcrystalline cellulose and water mixtures by controlled stress rheometry and the relationship to the production of spherical pellets by extrusion/spheronization.

The consistency of wet powder masses produced from two ratios (7:3 and 8:2) of alpha-lactose monohydrate (L) and microcrystalline cellulose (MCC) mixed with a range of water contents has been assessed with a parallel plate controlled stress rheometer. The range of water contents, which could be studied, was restricted to those, which could be extruded uniformly by a ram extruder. In the creep mode, the instantaneous compliance increased as the water content increased for both L:MCC ratios illustrating the increasing deformability of the mixtures with increasing water content. The derived apparent viscosity of the mixtures as a function of shear rate, increased as the water content decreased and the values for all the systems fell on a common line. This indicates that the measurements are providing a reliable assessment of the mixtures and that the change in water content and L:MCC ratio provides systems, whose change of viscosity with rate of shear is consistent at low rates of shear. The values of the storage and loss moduli obtained from oscillatory measurements, increased with a decrease in water content but this time the two ratios of L:MCC were not on a common line when related to the water content of the mixtures. There was a range of water levels over which both the values of the storage and loss moduli were approximately constant. This corresponded to the level of water, which produced the pellets of the smallest diameter and range of diameters and were of the most spherical shape when produced by a ram extruder and spheronization. For 8:2 L:MCC ratio, there appeared to be a value for both the storage and the loss moduli above which the wet mass could not produce good pellets. For the 7:3 L:MCC these limiting levels were not achieved before extrusion with steady state conditions could be maintained without the mass being too wet or too dry. Instead, there appeared to be minimum levels of the moduli required to ensure that the mixtures were able to produce good pellets.

A solitary neurofibroma of the palatine tonsil.

A rare case of a tonsillar neurofibroma in a 28-year-old Caucasian male is presented. Benign tumours of the tonsils are rare and of those reported, only a few are benign peripheral nerve sheath tumours (BPNSTs). This is the first report of a solitary neurofibroma of the palatine tonsil in the English literature.

Process control and scale-up of pharmaceutical wet granulation processes: a review.

In this paper the techniques for process control and scale-up of pharmaceutical wet granulation processes are reviewed. For wet granulation in high-shear mixers, specific methods based on the liquid saturation and the consistency of the wet mass are described. Both parameters can be used to quantify the deformability of the wet granules, and relate well with the particle size of the end granules. In practice, the power consumption of the high-shear mixer is used for the monitoring of the wet granulation process, whilst for scale-up, it is helpful to use the underlying relationship between power consumption and saturation level or wet mass consistency. In fluid bed granulation the granulation process is different and the moisture content in the bed is the key parameter to control. This can be monitored directly by near infrared probes or indirectly with temperature probes. As a large number of inter-related variables can be adjusted to modify the process, computerized techniques have become popular for fluid-bed process control--fuzzy logic, neural networks, and models based on experimental design techniques are several examples. In addition, engineering techniques based on particle size population balance modelling are under development for both fluid bed and high-shear granulation.

The changes in surface energetics with relative humidity of carbamazepine and paracetamol as measured by inverse gas chromatography.

The surface energetic parameters of carbamazepine and paracetamol have been studied using inverse gas chromatography modified to produce dry and ambient conditions within the column. The values of the dispersive component of the surface free energy (gamma(S)D) do not change significantly at the increased relative humidity. In contrast the specific component of the free energy of adsorption (-DeltaG(A)SP) as measured by polar probes, can either remain constant or decrease by up to 10%, depending on the material and the probe. This indicates that an increase in the relative humidity causes a decrease in the surface energetics of the powder surface. It is proposed that where the water molecules are adsorbing to the same sites as the polar probes, the interaction of these probes with the surface is decreased. To identify these sites, the preferential interaction of each probe, including water, with the drug molecule has been modelled.

Lessons for lecturers - ptosis and singultus. Private prescription: a thought-provoking tonic on the lighter side.

Please note that these are the personal opinions of the author and do not necessarily represent those of AstraZeneca.

Brittle-ductile transitions in sucrose and the influence of lateral stresses during compaction.

Sucrose, in a range of particle sizes, has been compacted to investigate both the effect of brittle-ductile transition and the effect of lateral stresses on the deformation stress as measured using Heckel plots. All particles with a diameter greater than 30 microm exhibited cracking in line with both theoretical predictions and literature data from hammer and ball milling. In addition, crack lengths in compressed particles examined microscopically were very similar to those predicted from the deformation stress, confirming the applicability of the model.