TCH-051 Validation of an Automated Compounder Set Up Once a Week For Parenteral Nutrition Solutions:

L Bouchoud, F Sadeghipour, S Fleury, P Bonnabry
2013 European journal of hospital pharmacy. Science and practice  
It is well known that the gustatory properties of a formulation strongly affect patient adherence to a treatment. However measuring these properties is highly subjective and difficult, especially for the paediatric population. The use of neuropsychophysiological indexes and covert behaviours in assessing the attractive properties of sensorial stimuli has a long tradition in the domain of affective neuroscience. Ways of measuring range from the use of autonomous nervous system activation
more » ... activation patterns, to features extracted from electroencephalographic activity or simple and discriminative reaction time tasks. These measurements provide alternative means for assessing the characteristics of commercial products, overcoming the limitations of self reporting-based research, namely social desirability, and for studying populations unable to provide usable verbal responses (e.g. children). Purpose To find out if this methodology can be used for evaluating the gustatory properties of formulations in order to enhance patient adherence. Materials and Methods Trimethoprim formulations were prepared using NF syrup. Flavour was added afterwards. Participants were stimulated with 3 different flavoured formulations (banana, red berry and neutral) for 10 seconds each while subjected to an EEG recording. The order of presentation was fully counter-balanced between subjects. Subjects rated the different solutions for palatability and intensity. Five seconds of the EEG response for each sample were converted to the frequency domain, and the log power and inter-hemispheric asymmetry were calculated for anterior, central and parietal electrodes. Different algorithms, combining different EEG features, were tested for predictive power regarding palatability and type of formula Results Theta inter-hemispheric activity at parietal electrodes predicted the behavioural assessment of palatability (R2 = 0.35). Moreover, the application of unsupervised learning methods, such as Support Vector Machines, on the log power at different bands from 0 to 12 Hz, could distinguish with up to 95.24% accuracy between flavoured and non-altered solutions. Conclusions This technology can be used in formulation studies that are attempting to enhance the organoleptic properties of a formulation.
doi:10.1136/ejhpharm-2013-000276.242 fatcat:l4aiss527rgw3jur6roe54y6pm